train_model.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The autoreload extension is already loaded. To reload it, use:\n",
      "  %reload_ext autoreload\n"
     ]
    }
   ],
   "source": [
    "# Import modules\n",
    "\n",
    "%matplotlib notebook\n",
    "\n",
    "from tensorflow import keras\n",
    "from tensorflow.keras.optimizers import Adam\n",
    "from tensorflow.keras.utils import plot_model\n",
    "from graph_model import CallBacks, DataGenerator, GraphNet\n",
    "\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import pandas as pd\n",
    "import os\n",
    "import pickle\n",
    "import shutil"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Define the data directories\n",
    "\n",
    "model_name = \"max-aggregate2_dropout=0.15+Gaussian\"\n",
    "\n",
    "cwd = os.getcwd()\n",
    "data_dir = os.path.join(cwd, \"data\")\n",
    "processed_waveform_dir = os.path.join(data_dir, \"waveforms_proc_broadband\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Data preparation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Fix random seed for reproducibility\n",
    "np.random.seed(0)\n",
    "\n",
    "# Data ranges (min/max)\n",
    "minlatitude = 32\n",
    "maxlatitude = 36\n",
    "minlongitude = -120\n",
    "maxlongitude = -116\n",
    "maxdepth = 30e3\n",
    "minmag = 3\n",
    "maxmag = 6\n",
    "\n",
    "# Fraction of the data to be used in training\n",
    "split = 0.8"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As part of the data preparation process, we scale the data within the range of $\\pm 1$, and split the data set into a training set and a validation set."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1377\n",
      "0.23999678936208627 50.291751151835314\n"
     ]
    },
    {
     "data": {
      "application/javascript": [
       "/* Put everything inside the global mpl namespace */\n",
       "window.mpl = {};\n",
       "\n",
       "\n",
       "mpl.get_websocket_type = function() {\n",
       "    if (typeof(WebSocket) !== 'undefined') {\n",
       "        return WebSocket;\n",
       "    } else if (typeof(MozWebSocket) !== 'undefined') {\n",
       "        return MozWebSocket;\n",
       "    } else {\n",
       "        alert('Your browser does not have WebSocket support. ' +\n",
       "              'Please try Chrome, Safari or Firefox ≥ 6. ' +\n",
       "              'Firefox 4 and 5 are also supported but you ' +\n",
       "              'have to enable WebSockets in about:config.');\n",
       "    };\n",
       "}\n",
       "\n",
       "mpl.figure = function(figure_id, websocket, ondownload, parent_element) {\n",
       "    this.id = figure_id;\n",
       "\n",
       "    this.ws = websocket;\n",
       "\n",
       "    this.supports_binary = (this.ws.binaryType != undefined);\n",
       "\n",
       "    if (!this.supports_binary) {\n",
       "        var warnings = document.getElementById(\"mpl-warnings\");\n",
       "        if (warnings) {\n",
       "            warnings.style.display = 'block';\n",
       "            warnings.textContent = (\n",
       "                \"This browser does not support binary websocket messages. \" +\n",
       "                    \"Performance may be slow.\");\n",
       "        }\n",
       "    }\n",
       "\n",
       "    this.imageObj = new Image();\n",
       "\n",
       "    this.context = undefined;\n",
       "    this.message = undefined;\n",
       "    this.canvas = undefined;\n",
       "    this.rubberband_canvas = undefined;\n",
       "    this.rubberband_context = undefined;\n",
       "    this.format_dropdown = undefined;\n",
       "\n",
       "    this.image_mode = 'full';\n",
       "\n",
       "    this.root = $('<div/>');\n",
       "    this._root_extra_style(this.root)\n",
       "    this.root.attr('style', 'display: inline-block');\n",
       "\n",
       "    $(parent_element).append(this.root);\n",
       "\n",
       "    this._init_header(this);\n",
       "    this._init_canvas(this);\n",
       "    this._init_toolbar(this);\n",
       "\n",
       "    var fig = this;\n",
       "\n",
       "    this.waiting = false;\n",
       "\n",
       "    this.ws.onopen =  function () {\n",
       "            fig.send_message(\"supports_binary\", {value: fig.supports_binary});\n",
       "            fig.send_message(\"send_image_mode\", {});\n",
       "            if (mpl.ratio != 1) {\n",
       "                fig.send_message(\"set_dpi_ratio\", {'dpi_ratio': mpl.ratio});\n",
       "            }\n",
       "            fig.send_message(\"refresh\", {});\n",
       "        }\n",
       "\n",
       "    this.imageObj.onload = function() {\n",
       "            if (fig.image_mode == 'full') {\n",
       "                // Full images could contain transparency (where diff images\n",
       "                // almost always do), so we need to clear the canvas so that\n",
       "                // there is no ghosting.\n",
       "                fig.context.clearRect(0, 0, fig.canvas.width, fig.canvas.height);\n",
       "            }\n",
       "            fig.context.drawImage(fig.imageObj, 0, 0);\n",
       "        };\n",
       "\n",
       "    this.imageObj.onunload = function() {\n",
       "        fig.ws.close();\n",
       "    }\n",
       "\n",
       "    this.ws.onmessage = this._make_on_message_function(this);\n",
       "\n",
       "    this.ondownload = ondownload;\n",
       "}\n",
       "\n",
       "mpl.figure.prototype._init_header = function() {\n",
       "    var titlebar = $(\n",
       "        '<div class=\"ui-dialog-titlebar ui-widget-header ui-corner-all ' +\n",
       "        'ui-helper-clearfix\"/>');\n",
       "    var titletext = $(\n",
       "        '<div class=\"ui-dialog-title\" style=\"width: 100%; ' +\n",
       "        'text-align: center; padding: 3px;\"/>');\n",
       "    titlebar.append(titletext)\n",
       "    this.root.append(titlebar);\n",
       "    this.header = titletext[0];\n",
       "}\n",
       "\n",
       "\n",
       "\n",
       "mpl.figure.prototype._canvas_extra_style = function(canvas_div) {\n",
       "\n",
       "}\n",
       "\n",
       "\n",
       "mpl.figure.prototype._root_extra_style = function(canvas_div) {\n",
       "\n",
       "}\n",
       "\n",
       "mpl.figure.prototype._init_canvas = function() {\n",
       "    var fig = this;\n",
       "\n",
       "    var canvas_div = $('<div/>');\n",
       "\n",
       "    canvas_div.attr('style', 'position: relative; clear: both; outline: 0');\n",
       "\n",
       "    function canvas_keyboard_event(event) {\n",
       "        return fig.key_event(event, event['data']);\n",
       "    }\n",
       "\n",
       "    canvas_div.keydown('key_press', canvas_keyboard_event);\n",
       "    canvas_div.keyup('key_release', canvas_keyboard_event);\n",
       "    this.canvas_div = canvas_div\n",
       "    this._canvas_extra_style(canvas_div)\n",
       "    this.root.append(canvas_div);\n",
       "\n",
       "    var canvas = $('<canvas/>');\n",
       "    canvas.addClass('mpl-canvas');\n",
       "    canvas.attr('style', \"left: 0; top: 0; z-index: 0; outline: 0\")\n",
       "\n",
       "    this.canvas = canvas[0];\n",
       "    this.context = canvas[0].getContext(\"2d\");\n",
       "\n",
       "    var backingStore = this.context.backingStorePixelRatio ||\n",
       "\tthis.context.webkitBackingStorePixelRatio ||\n",
       "\tthis.context.mozBackingStorePixelRatio ||\n",
       "\tthis.context.msBackingStorePixelRatio ||\n",
       "\tthis.context.oBackingStorePixelRatio ||\n",
       "\tthis.context.backingStorePixelRatio || 1;\n",
       "\n",
       "    mpl.ratio = (window.devicePixelRatio || 1) / backingStore;\n",
       "\n",
       "    var rubberband = $('<canvas/>');\n",
       "    rubberband.attr('style', \"position: absolute; left: 0; top: 0; z-index: 1;\")\n",
       "\n",
       "    var pass_mouse_events = true;\n",
       "\n",
       "    canvas_div.resizable({\n",
       "        start: function(event, ui) {\n",
       "            pass_mouse_events = false;\n",
       "        },\n",
       "        resize: function(event, ui) {\n",
       "            fig.request_resize(ui.size.width, ui.size.height);\n",
       "        },\n",
       "        stop: function(event, ui) {\n",
       "            pass_mouse_events = true;\n",
       "            fig.request_resize(ui.size.width, ui.size.height);\n",
       "        },\n",
       "    });\n",
       "\n",
       "    function mouse_event_fn(event) {\n",
       "        if (pass_mouse_events)\n",
       "            return fig.mouse_event(event, event['data']);\n",
       "    }\n",
       "\n",
       "    rubberband.mousedown('button_press', mouse_event_fn);\n",
       "    rubberband.mouseup('button_release', mouse_event_fn);\n",
       "    // Throttle sequential mouse events to 1 every 20ms.\n",
       "    rubberband.mousemove('motion_notify', mouse_event_fn);\n",
       "\n",
       "    rubberband.mouseenter('figure_enter', mouse_event_fn);\n",
       "    rubberband.mouseleave('figure_leave', mouse_event_fn);\n",
       "\n",
       "    canvas_div.on(\"wheel\", function (event) {\n",
       "        event = event.originalEvent;\n",
       "        event['data'] = 'scroll'\n",
       "        if (event.deltaY < 0) {\n",
       "            event.step = 1;\n",
       "        } else {\n",
       "            event.step = -1;\n",
       "        }\n",
       "        mouse_event_fn(event);\n",
       "    });\n",
       "\n",
       "    canvas_div.append(canvas);\n",
       "    canvas_div.append(rubberband);\n",
       "\n",
       "    this.rubberband = rubberband;\n",
       "    this.rubberband_canvas = rubberband[0];\n",
       "    this.rubberband_context = rubberband[0].getContext(\"2d\");\n",
       "    this.rubberband_context.strokeStyle = \"#000000\";\n",
       "\n",
       "    this._resize_canvas = function(width, height) {\n",
       "        // Keep the size of the canvas, canvas container, and rubber band\n",
       "        // canvas in synch.\n",
       "        canvas_div.css('width', width)\n",
       "        canvas_div.css('height', height)\n",
       "\n",
       "        canvas.attr('width', width * mpl.ratio);\n",
       "        canvas.attr('height', height * mpl.ratio);\n",
       "        canvas.attr('style', 'width: ' + width + 'px; height: ' + height + 'px;');\n",
       "\n",
       "        rubberband.attr('width', width);\n",
       "        rubberband.attr('height', height);\n",
       "    }\n",
       "\n",
       "    // Set the figure to an initial 600x600px, this will subsequently be updated\n",
       "    // upon first draw.\n",
       "    this._resize_canvas(600, 600);\n",
       "\n",
       "    // Disable right mouse context menu.\n",
       "    $(this.rubberband_canvas).bind(\"contextmenu\",function(e){\n",
       "        return false;\n",
       "    });\n",
       "\n",
       "    function set_focus () {\n",
       "        canvas.focus();\n",
       "        canvas_div.focus();\n",
       "    }\n",
       "\n",
       "    window.setTimeout(set_focus, 100);\n",
       "}\n",
       "\n",
       "mpl.figure.prototype._init_toolbar = function() {\n",
       "    var fig = this;\n",
       "\n",
       "    var nav_element = $('<div/>');\n",
       "    nav_element.attr('style', 'width: 100%');\n",
       "    this.root.append(nav_element);\n",
       "\n",
       "    // Define a callback function for later on.\n",
       "    function toolbar_event(event) {\n",
       "        return fig.toolbar_button_onclick(event['data']);\n",
       "    }\n",
       "    function toolbar_mouse_event(event) {\n",
       "        return fig.toolbar_button_onmouseover(event['data']);\n",
       "    }\n",
       "\n",
       "    for(var toolbar_ind in mpl.toolbar_items) {\n",
       "        var name = mpl.toolbar_items[toolbar_ind][0];\n",
       "        var tooltip = mpl.toolbar_items[toolbar_ind][1];\n",
       "        var image = mpl.toolbar_items[toolbar_ind][2];\n",
       "        var method_name = mpl.toolbar_items[toolbar_ind][3];\n",
       "\n",
       "        if (!name) {\n",
       "            // put a spacer in here.\n",
       "            continue;\n",
       "        }\n",
       "        var button = $('<button/>');\n",
       "        button.addClass('ui-button ui-widget ui-state-default ui-corner-all ' +\n",
       "                        'ui-button-icon-only');\n",
       "        button.attr('role', 'button');\n",
       "        button.attr('aria-disabled', 'false');\n",
       "        button.click(method_name, toolbar_event);\n",
       "        button.mouseover(tooltip, toolbar_mouse_event);\n",
       "\n",
       "        var icon_img = $('<span/>');\n",
       "        icon_img.addClass('ui-button-icon-primary ui-icon');\n",
       "        icon_img.addClass(image);\n",
       "        icon_img.addClass('ui-corner-all');\n",
       "\n",
       "        var tooltip_span = $('<span/>');\n",
       "        tooltip_span.addClass('ui-button-text');\n",
       "        tooltip_span.html(tooltip);\n",
       "\n",
       "        button.append(icon_img);\n",
       "        button.append(tooltip_span);\n",
       "\n",
       "        nav_element.append(button);\n",
       "    }\n",
       "\n",
       "    var fmt_picker_span = $('<span/>');\n",
       "\n",
       "    var fmt_picker = $('<select/>');\n",
       "    fmt_picker.addClass('mpl-toolbar-option ui-widget ui-widget-content');\n",
       "    fmt_picker_span.append(fmt_picker);\n",
       "    nav_element.append(fmt_picker_span);\n",
       "    this.format_dropdown = fmt_picker[0];\n",
       "\n",
       "    for (var ind in mpl.extensions) {\n",
       "        var fmt = mpl.extensions[ind];\n",
       "        var option = $(\n",
       "            '<option/>', {selected: fmt === mpl.default_extension}).html(fmt);\n",
       "        fmt_picker.append(option);\n",
       "    }\n",
       "\n",
       "    // Add hover states to the ui-buttons\n",
       "    $( \".ui-button\" ).hover(\n",
       "        function() { $(this).addClass(\"ui-state-hover\");},\n",
       "        function() { $(this).removeClass(\"ui-state-hover\");}\n",
       "    );\n",
       "\n",
       "    var status_bar = $('<span class=\"mpl-message\"/>');\n",
       "    nav_element.append(status_bar);\n",
       "    this.message = status_bar[0];\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.request_resize = function(x_pixels, y_pixels) {\n",
       "    // Request matplotlib to resize the figure. Matplotlib will then trigger a resize in the client,\n",
       "    // which will in turn request a refresh of the image.\n",
       "    this.send_message('resize', {'width': x_pixels, 'height': y_pixels});\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.send_message = function(type, properties) {\n",
       "    properties['type'] = type;\n",
       "    properties['figure_id'] = this.id;\n",
       "    this.ws.send(JSON.stringify(properties));\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.send_draw_message = function() {\n",
       "    if (!this.waiting) {\n",
       "        this.waiting = true;\n",
       "        this.ws.send(JSON.stringify({type: \"draw\", figure_id: this.id}));\n",
       "    }\n",
       "}\n",
       "\n",
       "\n",
       "mpl.figure.prototype.handle_save = function(fig, msg) {\n",
       "    var format_dropdown = fig.format_dropdown;\n",
       "    var format = format_dropdown.options[format_dropdown.selectedIndex].value;\n",
       "    fig.ondownload(fig, format);\n",
       "}\n",
       "\n",
       "\n",
       "mpl.figure.prototype.handle_resize = function(fig, msg) {\n",
       "    var size = msg['size'];\n",
       "    if (size[0] != fig.canvas.width || size[1] != fig.canvas.height) {\n",
       "        fig._resize_canvas(size[0], size[1]);\n",
       "        fig.send_message(\"refresh\", {});\n",
       "    };\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.handle_rubberband = function(fig, msg) {\n",
       "    var x0 = msg['x0'] / mpl.ratio;\n",
       "    var y0 = (fig.canvas.height - msg['y0']) / mpl.ratio;\n",
       "    var x1 = msg['x1'] / mpl.ratio;\n",
       "    var y1 = (fig.canvas.height - msg['y1']) / mpl.ratio;\n",
       "    x0 = Math.floor(x0) + 0.5;\n",
       "    y0 = Math.floor(y0) + 0.5;\n",
       "    x1 = Math.floor(x1) + 0.5;\n",
       "    y1 = Math.floor(y1) + 0.5;\n",
       "    var min_x = Math.min(x0, x1);\n",
       "    var min_y = Math.min(y0, y1);\n",
       "    var width = Math.abs(x1 - x0);\n",
       "    var height = Math.abs(y1 - y0);\n",
       "\n",
       "    fig.rubberband_context.clearRect(\n",
       "        0, 0, fig.canvas.width / mpl.ratio, fig.canvas.height / mpl.ratio);\n",
       "\n",
       "    fig.rubberband_context.strokeRect(min_x, min_y, width, height);\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.handle_figure_label = function(fig, msg) {\n",
       "    // Updates the figure title.\n",
       "    fig.header.textContent = msg['label'];\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.handle_cursor = function(fig, msg) {\n",
       "    var cursor = msg['cursor'];\n",
       "    switch(cursor)\n",
       "    {\n",
       "    case 0:\n",
       "        cursor = 'pointer';\n",
       "        break;\n",
       "    case 1:\n",
       "        cursor = 'default';\n",
       "        break;\n",
       "    case 2:\n",
       "        cursor = 'crosshair';\n",
       "        break;\n",
       "    case 3:\n",
       "        cursor = 'move';\n",
       "        break;\n",
       "    }\n",
       "    fig.rubberband_canvas.style.cursor = cursor;\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.handle_message = function(fig, msg) {\n",
       "    fig.message.textContent = msg['message'];\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.handle_draw = function(fig, msg) {\n",
       "    // Request the server to send over a new figure.\n",
       "    fig.send_draw_message();\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.handle_image_mode = function(fig, msg) {\n",
       "    fig.image_mode = msg['mode'];\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.updated_canvas_event = function() {\n",
       "    // Called whenever the canvas gets updated.\n",
       "    this.send_message(\"ack\", {});\n",
       "}\n",
       "\n",
       "// A function to construct a web socket function for onmessage handling.\n",
       "// Called in the figure constructor.\n",
       "mpl.figure.prototype._make_on_message_function = function(fig) {\n",
       "    return function socket_on_message(evt) {\n",
       "        if (evt.data instanceof Blob) {\n",
       "            /* FIXME: We get \"Resource interpreted as Image but\n",
       "             * transferred with MIME type text/plain:\" errors on\n",
       "             * Chrome.  But how to set the MIME type?  It doesn't seem\n",
       "             * to be part of the websocket stream */\n",
       "            evt.data.type = \"image/png\";\n",
       "\n",
       "            /* Free the memory for the previous frames */\n",
       "            if (fig.imageObj.src) {\n",
       "                (window.URL || window.webkitURL).revokeObjectURL(\n",
       "                    fig.imageObj.src);\n",
       "            }\n",
       "\n",
       "            fig.imageObj.src = (window.URL || window.webkitURL).createObjectURL(\n",
       "                evt.data);\n",
       "            fig.updated_canvas_event();\n",
       "            fig.waiting = false;\n",
       "            return;\n",
       "        }\n",
       "        else if (typeof evt.data === 'string' && evt.data.slice(0, 21) == \"data:image/png;base64\") {\n",
       "            fig.imageObj.src = evt.data;\n",
       "            fig.updated_canvas_event();\n",
       "            fig.waiting = false;\n",
       "            return;\n",
       "        }\n",
       "\n",
       "        var msg = JSON.parse(evt.data);\n",
       "        var msg_type = msg['type'];\n",
       "\n",
       "        // Call the  \"handle_{type}\" callback, which takes\n",
       "        // the figure and JSON message as its only arguments.\n",
       "        try {\n",
       "            var callback = fig[\"handle_\" + msg_type];\n",
       "        } catch (e) {\n",
       "            console.log(\"No handler for the '\" + msg_type + \"' message type: \", msg);\n",
       "            return;\n",
       "        }\n",
       "\n",
       "        if (callback) {\n",
       "            try {\n",
       "                // console.log(\"Handling '\" + msg_type + \"' message: \", msg);\n",
       "                callback(fig, msg);\n",
       "            } catch (e) {\n",
       "                console.log(\"Exception inside the 'handler_\" + msg_type + \"' callback:\", e, e.stack, msg);\n",
       "            }\n",
       "        }\n",
       "    };\n",
       "}\n",
       "\n",
       "// from http://stackoverflow.com/questions/1114465/getting-mouse-location-in-canvas\n",
       "mpl.findpos = function(e) {\n",
       "    //this section is from http://www.quirksmode.org/js/events_properties.html\n",
       "    var targ;\n",
       "    if (!e)\n",
       "        e = window.event;\n",
       "    if (e.target)\n",
       "        targ = e.target;\n",
       "    else if (e.srcElement)\n",
       "        targ = e.srcElement;\n",
       "    if (targ.nodeType == 3) // defeat Safari bug\n",
       "        targ = targ.parentNode;\n",
       "\n",
       "    // jQuery normalizes the pageX and pageY\n",
       "    // pageX,Y are the mouse positions relative to the document\n",
       "    // offset() returns the position of the element relative to the document\n",
       "    var x = e.pageX - $(targ).offset().left;\n",
       "    var y = e.pageY - $(targ).offset().top;\n",
       "\n",
       "    return {\"x\": x, \"y\": y};\n",
       "};\n",
       "\n",
       "/*\n",
       " * return a copy of an object with only non-object keys\n",
       " * we need this to avoid circular references\n",
       " * http://stackoverflow.com/a/24161582/3208463\n",
       " */\n",
       "function simpleKeys (original) {\n",
       "  return Object.keys(original).reduce(function (obj, key) {\n",
       "    if (typeof original[key] !== 'object')\n",
       "        obj[key] = original[key]\n",
       "    return obj;\n",
       "  }, {});\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.mouse_event = function(event, name) {\n",
       "    var canvas_pos = mpl.findpos(event)\n",
       "\n",
       "    if (name === 'button_press')\n",
       "    {\n",
       "        this.canvas.focus();\n",
       "        this.canvas_div.focus();\n",
       "    }\n",
       "\n",
       "    var x = canvas_pos.x * mpl.ratio;\n",
       "    var y = canvas_pos.y * mpl.ratio;\n",
       "\n",
       "    this.send_message(name, {x: x, y: y, button: event.button,\n",
       "                             step: event.step,\n",
       "                             guiEvent: simpleKeys(event)});\n",
       "\n",
       "    /* This prevents the web browser from automatically changing to\n",
       "     * the text insertion cursor when the button is pressed.  We want\n",
       "     * to control all of the cursor setting manually through the\n",
       "     * 'cursor' event from matplotlib */\n",
       "    event.preventDefault();\n",
       "    return false;\n",
       "}\n",
       "\n",
       "mpl.figure.prototype._key_event_extra = function(event, name) {\n",
       "    // Handle any extra behaviour associated with a key event\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.key_event = function(event, name) {\n",
       "\n",
       "    // Prevent repeat events\n",
       "    if (name == 'key_press')\n",
       "    {\n",
       "        if (event.which === this._key)\n",
       "            return;\n",
       "        else\n",
       "            this._key = event.which;\n",
       "    }\n",
       "    if (name == 'key_release')\n",
       "        this._key = null;\n",
       "\n",
       "    var value = '';\n",
       "    if (event.ctrlKey && event.which != 17)\n",
       "        value += \"ctrl+\";\n",
       "    if (event.altKey && event.which != 18)\n",
       "        value += \"alt+\";\n",
       "    if (event.shiftKey && event.which != 16)\n",
       "        value += \"shift+\";\n",
       "\n",
       "    value += 'k';\n",
       "    value += event.which.toString();\n",
       "\n",
       "    this._key_event_extra(event, name);\n",
       "\n",
       "    this.send_message(name, {key: value,\n",
       "                             guiEvent: simpleKeys(event)});\n",
       "    return false;\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.toolbar_button_onclick = function(name) {\n",
       "    if (name == 'download') {\n",
       "        this.handle_save(this, null);\n",
       "    } else {\n",
       "        this.send_message(\"toolbar_button\", {name: name});\n",
       "    }\n",
       "};\n",
       "\n",
       "mpl.figure.prototype.toolbar_button_onmouseover = function(tooltip) {\n",
       "    this.message.textContent = tooltip;\n",
       "};\n",
       "mpl.toolbar_items = [[\"Home\", \"Reset original view\", \"fa fa-home icon-home\", \"home\"], [\"Back\", \"Back to previous view\", \"fa fa-arrow-left icon-arrow-left\", \"back\"], [\"Forward\", \"Forward to next view\", \"fa fa-arrow-right icon-arrow-right\", \"forward\"], [\"\", \"\", \"\", \"\"], [\"Pan\", \"Pan axes with left mouse, zoom with right\", \"fa fa-arrows icon-move\", \"pan\"], [\"Zoom\", \"Zoom to rectangle\", \"fa fa-square-o icon-check-empty\", \"zoom\"], [\"\", \"\", \"\", \"\"], [\"Download\", \"Download plot\", \"fa fa-floppy-o icon-save\", \"download\"]];\n",
       "\n",
       "mpl.extensions = [\"eps\", \"jpeg\", \"pdf\", \"png\", \"ps\", \"raw\", \"svg\", \"tif\"];\n",
       "\n",
       "mpl.default_extension = \"png\";var comm_websocket_adapter = function(comm) {\n",
       "    // Create a \"websocket\"-like object which calls the given IPython comm\n",
       "    // object with the appropriate methods. Currently this is a non binary\n",
       "    // socket, so there is still some room for performance tuning.\n",
       "    var ws = {};\n",
       "\n",
       "    ws.close = function() {\n",
       "        comm.close()\n",
       "    };\n",
       "    ws.send = function(m) {\n",
       "        //console.log('sending', m);\n",
       "        comm.send(m);\n",
       "    };\n",
       "    // Register the callback with on_msg.\n",
       "    comm.on_msg(function(msg) {\n",
       "        //console.log('receiving', msg['content']['data'], msg);\n",
       "        // Pass the mpl event to the overridden (by mpl) onmessage function.\n",
       "        ws.onmessage(msg['content']['data'])\n",
       "    });\n",
       "    return ws;\n",
       "}\n",
       "\n",
       "mpl.mpl_figure_comm = function(comm, msg) {\n",
       "    // This is the function which gets called when the mpl process\n",
       "    // starts-up an IPython Comm through the \"matplotlib\" channel.\n",
       "\n",
       "    var id = msg.content.data.id;\n",
       "    // Get hold of the div created by the display call when the Comm\n",
       "    // socket was opened in Python.\n",
       "    var element = $(\"#\" + id);\n",
       "    var ws_proxy = comm_websocket_adapter(comm)\n",
       "\n",
       "    function ondownload(figure, format) {\n",
       "        window.open(figure.imageObj.src);\n",
       "    }\n",
       "\n",
       "    var fig = new mpl.figure(id, ws_proxy,\n",
       "                           ondownload,\n",
       "                           element.get(0));\n",
       "\n",
       "    // Call onopen now - mpl needs it, as it is assuming we've passed it a real\n",
       "    // web socket which is closed, not our websocket->open comm proxy.\n",
       "    ws_proxy.onopen();\n",
       "\n",
       "    fig.parent_element = element.get(0);\n",
       "    fig.cell_info = mpl.find_output_cell(\"<div id='\" + id + \"'></div>\");\n",
       "    if (!fig.cell_info) {\n",
       "        console.error(\"Failed to find cell for figure\", id, fig);\n",
       "        return;\n",
       "    }\n",
       "\n",
       "    var output_index = fig.cell_info[2]\n",
       "    var cell = fig.cell_info[0];\n",
       "\n",
       "};\n",
       "\n",
       "mpl.figure.prototype.handle_close = function(fig, msg) {\n",
       "    var width = fig.canvas.width/mpl.ratio\n",
       "    fig.root.unbind('remove')\n",
       "\n",
       "    // Update the output cell to use the data from the current canvas.\n",
       "    fig.push_to_output();\n",
       "    var dataURL = fig.canvas.toDataURL();\n",
       "    // Re-enable the keyboard manager in IPython - without this line, in FF,\n",
       "    // the notebook keyboard shortcuts fail.\n",
       "    IPython.keyboard_manager.enable()\n",
       "    $(fig.parent_element).html('<img src=\"' + dataURL + '\" width=\"' + width + '\">');\n",
       "    fig.close_ws(fig, msg);\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.close_ws = function(fig, msg){\n",
       "    fig.send_message('closing', msg);\n",
       "    // fig.ws.close()\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.push_to_output = function(remove_interactive) {\n",
       "    // Turn the data on the canvas into data in the output cell.\n",
       "    var width = this.canvas.width/mpl.ratio\n",
       "    var dataURL = this.canvas.toDataURL();\n",
       "    this.cell_info[1]['text/html'] = '<img src=\"' + dataURL + '\" width=\"' + width + '\">';\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.updated_canvas_event = function() {\n",
       "    // Tell IPython that the notebook contents must change.\n",
       "    IPython.notebook.set_dirty(true);\n",
       "    this.send_message(\"ack\", {});\n",
       "    var fig = this;\n",
       "    // Wait a second, then push the new image to the DOM so\n",
       "    // that it is saved nicely (might be nice to debounce this).\n",
       "    setTimeout(function () { fig.push_to_output() }, 1000);\n",
       "}\n",
       "\n",
       "mpl.figure.prototype._init_toolbar = function() {\n",
       "    var fig = this;\n",
       "\n",
       "    var nav_element = $('<div/>');\n",
       "    nav_element.attr('style', 'width: 100%');\n",
       "    this.root.append(nav_element);\n",
       "\n",
       "    // Define a callback function for later on.\n",
       "    function toolbar_event(event) {\n",
       "        return fig.toolbar_button_onclick(event['data']);\n",
       "    }\n",
       "    function toolbar_mouse_event(event) {\n",
       "        return fig.toolbar_button_onmouseover(event['data']);\n",
       "    }\n",
       "\n",
       "    for(var toolbar_ind in mpl.toolbar_items){\n",
       "        var name = mpl.toolbar_items[toolbar_ind][0];\n",
       "        var tooltip = mpl.toolbar_items[toolbar_ind][1];\n",
       "        var image = mpl.toolbar_items[toolbar_ind][2];\n",
       "        var method_name = mpl.toolbar_items[toolbar_ind][3];\n",
       "\n",
       "        if (!name) { continue; };\n",
       "\n",
       "        var button = $('<button class=\"btn btn-default\" href=\"#\" title=\"' + name + '\"><i class=\"fa ' + image + ' fa-lg\"></i></button>');\n",
       "        button.click(method_name, toolbar_event);\n",
       "        button.mouseover(tooltip, toolbar_mouse_event);\n",
       "        nav_element.append(button);\n",
       "    }\n",
       "\n",
       "    // Add the status bar.\n",
       "    var status_bar = $('<span class=\"mpl-message\" style=\"text-align:right; float: right;\"/>');\n",
       "    nav_element.append(status_bar);\n",
       "    this.message = status_bar[0];\n",
       "\n",
       "    // Add the close button to the window.\n",
       "    var buttongrp = $('<div class=\"btn-group inline pull-right\"></div>');\n",
       "    var button = $('<button class=\"btn btn-mini btn-primary\" href=\"#\" title=\"Stop Interaction\"><i class=\"fa fa-power-off icon-remove icon-large\"></i></button>');\n",
       "    button.click(function (evt) { fig.handle_close(fig, {}); } );\n",
       "    button.mouseover('Stop Interaction', toolbar_mouse_event);\n",
       "    buttongrp.append(button);\n",
       "    var titlebar = this.root.find($('.ui-dialog-titlebar'));\n",
       "    titlebar.prepend(buttongrp);\n",
       "}\n",
       "\n",
       "mpl.figure.prototype._root_extra_style = function(el){\n",
       "    var fig = this\n",
       "    el.on(\"remove\", function(){\n",
       "\tfig.close_ws(fig, {});\n",
       "    });\n",
       "}\n",
       "\n",
       "mpl.figure.prototype._canvas_extra_style = function(el){\n",
       "    // this is important to make the div 'focusable\n",
       "    el.attr('tabindex', 0)\n",
       "    // reach out to IPython and tell the keyboard manager to turn it's self\n",
       "    // off when our div gets focus\n",
       "\n",
       "    // location in version 3\n",
       "    if (IPython.notebook.keyboard_manager) {\n",
       "        IPython.notebook.keyboard_manager.register_events(el);\n",
       "    }\n",
       "    else {\n",
       "        // location in version 2\n",
       "        IPython.keyboard_manager.register_events(el);\n",
       "    }\n",
       "\n",
       "}\n",
       "\n",
       "mpl.figure.prototype._key_event_extra = function(event, name) {\n",
       "    var manager = IPython.notebook.keyboard_manager;\n",
       "    if (!manager)\n",
       "        manager = IPython.keyboard_manager;\n",
       "\n",
       "    // Check for shift+enter\n",
       "    if (event.shiftKey && event.which == 13) {\n",
       "        this.canvas_div.blur();\n",
       "        // select the cell after this one\n",
       "        var index = IPython.notebook.find_cell_index(this.cell_info[0]);\n",
       "        IPython.notebook.select(index + 1);\n",
       "    }\n",
       "}\n",
       "\n",
       "mpl.figure.prototype.handle_save = function(fig, msg) {\n",
       "    fig.ondownload(fig, null);\n",
       "}\n",
       "\n",
       "\n",
       "mpl.find_output_cell = function(html_output) {\n",
       "    // Return the cell and output element which can be found *uniquely* in the notebook.\n",
       "    // Note - this is a bit hacky, but it is done because the \"notebook_saving.Notebook\"\n",
       "    // IPython event is triggered only after the cells have been serialised, which for\n",
       "    // our purposes (turning an active figure into a static one), is too late.\n",
       "    var cells = IPython.notebook.get_cells();\n",
       "    var ncells = cells.length;\n",
       "    for (var i=0; i<ncells; i++) {\n",
       "        var cell = cells[i];\n",
       "        if (cell.cell_type === 'code'){\n",
       "            for (var j=0; j<cell.output_area.outputs.length; j++) {\n",
       "                var data = cell.output_area.outputs[j];\n",
       "                if (data.data) {\n",
       "                    // IPython >= 3 moved mimebundle to data attribute of output\n",
       "                    data = data.data;\n",
       "                }\n",
       "                if (data['text/html'] == html_output) {\n",
       "                    return [cell, data, j];\n",
       "                }\n",
       "            }\n",
       "        }\n",
       "    }\n",
       "}\n",
       "\n",
       "// Register the function which deals with the matplotlib target/channel.\n",
       "// The kernel may be null if the page has been refreshed.\n",
       "if (IPython.notebook.kernel != null) {\n",
       "    IPython.notebook.kernel.comm_manager.register_target('matplotlib', mpl.mpl_figure_comm);\n",
       "}\n"
      ],
      "text/plain": [
       "<IPython.core.display.Javascript object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<img src=\"data:image/png;base64,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\" width=\"900\">"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Get the event catalogue\n",
    "catalogue = pd.read_csv(os.path.join(data_dir, \"catalogue.csv\"))[[\"lat\", \"lon\", \"depth\", \"mag\"]]\n",
    "\n",
    "# Check which events have data\n",
    "for i, event in catalogue.iterrows():\n",
    "    # Event data file\n",
    "    cat_file = os.path.join(processed_waveform_dir, \"%d.npy\" % i)\n",
    "    # If file not exists: remove event from catalogue\n",
    "    if not os.path.isfile(cat_file):\n",
    "        catalogue.drop(index=i, inplace=True)\n",
    "\n",
    "# Print number of events\n",
    "print(len(catalogue))        \n",
    "\n",
    "# Event identifiers\n",
    "ids = catalogue.index.values\n",
    "# Uniform weights\n",
    "weights = np.ones((len(ids), 1))\n",
    "# Event lat/lon/depth/magnitude\n",
    "catalogue = catalogue.values\n",
    "\n",
    "# Scale data\n",
    "catalogue[:, 0] = (catalogue[:, 0] - minlatitude) / (maxlatitude - minlatitude)\n",
    "catalogue[:, 1] = (catalogue[:, 1] - minlongitude) / (maxlongitude - minlongitude)\n",
    "catalogue[:, 2] = catalogue[:, 2] / maxdepth\n",
    "catalogue[:, 3] = (catalogue[:, 3] - minmag) / (maxmag - minmag)\n",
    "\n",
    "catalogue = (catalogue - 0.5) * 2\n",
    "\n",
    "# Concatenate identifiers and weights to event data\n",
    "catalogue = np.concatenate([ids.reshape(-1, 1), weights.reshape(-1, 1), catalogue], axis=1)\n",
    "\n",
    "# Randomly split events into train and validation sets\n",
    "inds = np.arange(catalogue.shape[0])\n",
    "np.random.shuffle(inds)\n",
    "N_split = int(split * catalogue.shape[0])\n",
    "\n",
    "train_inds = inds[:N_split]\n",
    "test_inds = inds[N_split:]\n",
    "\n",
    "# Split catalogue\n",
    "train_catalogue = catalogue[train_inds]\n",
    "test_catalogue = catalogue[test_inds]\n",
    "\n",
    "# Check data distributions to ensure that train and validation sets are similar!\n",
    "fig, axes = plt.subplots(nrows=2, ncols=2, figsize=(9, 7))\n",
    "axes = axes.ravel()\n",
    "\n",
    "for i in range(4):\n",
    "    ax = axes[i]\n",
    "    ax.hist(catalogue[:, i+2], bins=20, density=False)\n",
    "    ax.hist(train_catalogue[:, i+2], bins=20, alpha=0.5, density=False)\n",
    "    ax.hist(test_catalogue[:, i+2], bins=20, alpha=0.5, density=False)\n",
    "\n",
    "plt.tight_layout()    \n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Read seismic station information\n",
    "stations = pd.read_csv(os.path.join(data_dir, \"stations.csv\"))[[\"code\", \"lat\", \"lon\"]]\n",
    "# Scale station geographic locations\n",
    "stations[\"lat\"] = ((stations[\"lat\"] - minlatitude) / (maxlatitude - minlatitude) - 0.5) * 2\n",
    "stations[\"lon\"] = ((stations[\"lon\"] - minlongitude) / (maxlongitude - minlongitude) - 0.5) * 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Open the event-station lookup file\n",
    "lookup_file = os.path.join(data_dir, \"catalogue_station_lookup_final.pickle\")\n",
    "\n",
    "with open(lookup_file, \"rb\") as f:\n",
    "    lookup = pickle.load(f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Number of stations per sample\n",
    "N_sub = 50\n",
    "# Number of time sample points\n",
    "N_t = 2048\n",
    "\n",
    "# Instantiate train data generator\n",
    "train_generator = DataGenerator(\n",
    "    data_dir=processed_waveform_dir,\n",
    "    catalogue=train_catalogue,\n",
    "    stations=stations,\n",
    "    lookup=lookup,\n",
    "    N_sub=N_sub,\n",
    "    N_t=N_t,\n",
    "    batch_size=32,\n",
    ")\n",
    "\n",
    "# Instantiate validation data generator\n",
    "test_generator = DataGenerator(\n",
    "    data_dir=processed_waveform_dir,\n",
    "    catalogue=test_catalogue,\n",
    "    stations=stations,\n",
    "    lookup=lookup,\n",
    "    N_sub=N_sub,\n",
    "    N_t=N_t,\n",
    "    batch_size=32,\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Model construction"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "\"\"\" Define Tensorflow/Keras callbacks \"\"\"\n",
    "\n",
    "# Tensorboard log directory\n",
    "logdir = os.path.join(\"logs\", model_name)\n",
    "if os.path.isdir(logdir):\n",
    "    shutil.rmtree(logdir)\n",
    "\n",
    "# Save model directory\n",
    "savedir = os.path.join(\"save\", model_name)\n",
    "if not os.path.isdir(savedir):\n",
    "    os.makedirs(savedir)\n",
    "\n",
    "# Best and last model save files\n",
    "savefile_best = \"best-model.h5\"\n",
    "savefile_last = \"last-model.h5\"\n",
    "\n",
    "# Instantiate callbacks\n",
    "tensorboard_callback = CallBacks.tensorboard(logdir)\n",
    "checkpoint_callback_best = CallBacks.checkpoint(os.path.join(savedir, savefile_best), best=True)\n",
    "checkpoint_callback_last = CallBacks.checkpoint(os.path.join(savedir, savefile_last), best=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"model_3\"\n",
      "__________________________________________________________________________________________________\n",
      "Layer (type)                    Output Shape         Param #     Connected to                     \n",
      "==================================================================================================\n",
      "input_10 (InputLayer)           [(None, 50, 2048, 3) 0                                            \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_51 (Conv2D)              (None, 50, 2048, 4)  64          input_10[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "activation_51 (Activation)      (None, 50, 2048, 4)  0           conv2d_51[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "gaussian_dropout_54 (GaussianDr (None, 50, 2048, 4)  0           activation_51[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_52 (Conv2D)              (None, 50, 2048, 4)  84          gaussian_dropout_54[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "activation_52 (Activation)      (None, 50, 2048, 4)  0           conv2d_52[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "gaussian_dropout_55 (GaussianDr (None, 50, 2048, 4)  0           activation_52[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_53 (Conv2D)              (None, 50, 2048, 4)  84          gaussian_dropout_55[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "activation_53 (Activation)      (None, 50, 2048, 4)  0           conv2d_53[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "gaussian_dropout_56 (GaussianDr (None, 50, 2048, 4)  0           activation_53[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling2d_12 (MaxPooling2D) (None, 50, 512, 4)   0           gaussian_dropout_56[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_54 (Conv2D)              (None, 50, 512, 8)   168         max_pooling2d_12[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "activation_54 (Activation)      (None, 50, 512, 8)   0           conv2d_54[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "gaussian_dropout_57 (GaussianDr (None, 50, 512, 8)   0           activation_54[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_55 (Conv2D)              (None, 50, 512, 8)   328         gaussian_dropout_57[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "activation_55 (Activation)      (None, 50, 512, 8)   0           conv2d_55[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "gaussian_dropout_58 (GaussianDr (None, 50, 512, 8)   0           activation_55[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_56 (Conv2D)              (None, 50, 512, 8)   328         gaussian_dropout_58[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "activation_56 (Activation)      (None, 50, 512, 8)   0           conv2d_56[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "gaussian_dropout_59 (GaussianDr (None, 50, 512, 8)   0           activation_56[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling2d_13 (MaxPooling2D) (None, 50, 128, 8)   0           gaussian_dropout_59[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_57 (Conv2D)              (None, 50, 128, 16)  656         max_pooling2d_13[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "activation_57 (Activation)      (None, 50, 128, 16)  0           conv2d_57[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "gaussian_dropout_60 (GaussianDr (None, 50, 128, 16)  0           activation_57[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_58 (Conv2D)              (None, 50, 128, 16)  1296        gaussian_dropout_60[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "activation_58 (Activation)      (None, 50, 128, 16)  0           conv2d_58[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "gaussian_dropout_61 (GaussianDr (None, 50, 128, 16)  0           activation_58[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_59 (Conv2D)              (None, 50, 128, 16)  1296        gaussian_dropout_61[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "activation_59 (Activation)      (None, 50, 128, 16)  0           conv2d_59[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "gaussian_dropout_62 (GaussianDr (None, 50, 128, 16)  0           activation_59[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling2d_14 (MaxPooling2D) (None, 50, 32, 16)   0           gaussian_dropout_62[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_60 (Conv2D)              (None, 50, 32, 32)   2592        max_pooling2d_14[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "activation_60 (Activation)      (None, 50, 32, 32)   0           conv2d_60[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "gaussian_dropout_63 (GaussianDr (None, 50, 32, 32)   0           activation_60[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_61 (Conv2D)              (None, 50, 32, 32)   5152        gaussian_dropout_63[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "activation_61 (Activation)      (None, 50, 32, 32)   0           conv2d_61[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "gaussian_dropout_64 (GaussianDr (None, 50, 32, 32)   0           activation_61[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_62 (Conv2D)              (None, 50, 32, 32)   5152        gaussian_dropout_64[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "activation_62 (Activation)      (None, 50, 32, 32)   0           conv2d_62[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "gaussian_dropout_65 (GaussianDr (None, 50, 32, 32)   0           activation_62[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling2d_15 (MaxPooling2D) (None, 50, 8, 32)    0           gaussian_dropout_65[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_63 (Conv2D)              (None, 50, 8, 64)    10304       max_pooling2d_15[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "activation_63 (Activation)      (None, 50, 8, 64)    0           conv2d_63[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "gaussian_dropout_66 (GaussianDr (None, 50, 8, 64)    0           activation_63[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_64 (Conv2D)              (None, 50, 8, 64)    20544       gaussian_dropout_66[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "activation_64 (Activation)      (None, 50, 8, 64)    0           conv2d_64[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "gaussian_dropout_67 (GaussianDr (None, 50, 8, 64)    0           activation_64[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_65 (Conv2D)              (None, 50, 8, 64)    20544       gaussian_dropout_67[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "activation_65 (Activation)      (None, 50, 8, 64)    0           conv2d_65[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "lambda_6 (Lambda)               (None, 50, 1, 64)    0           activation_65[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "gaussian_dropout_68 (GaussianDr (None, 50, 1, 64)    0           lambda_6[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "input_11 (InputLayer)           [(None, 50, 1, 3)]   0                                            \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_3 (Concatenate)     (None, 50, 1, 67)    0           gaussian_dropout_68[0][0]        \n",
      "                                                                 input_11[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_66 (Conv2D)              (None, 50, 1, 128)   8704        concatenate_3[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_66 (Activation)      (None, 50, 1, 128)   0           conv2d_66[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "input_12 (InputLayer)           [(None, 50)]         0                                            \n",
      "__________________________________________________________________________________________________\n",
      "gaussian_dropout_69 (GaussianDr (None, 50, 1, 128)   0           activation_66[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "repeat_vector_3 (RepeatVector)  (None, 128, 50)      0           input_12[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_67 (Conv2D)              (None, 50, 1, 128)   16512       gaussian_dropout_69[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "permute_3 (Permute)             (None, 50, 128)      0           repeat_vector_3[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "activation_67 (Activation)      (None, 50, 1, 128)   0           conv2d_67[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "tf_op_layer_ExpandDims_3 (Tenso [(None, 50, 1, 128)] 0           permute_3[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "tf_op_layer_mul_3 (TensorFlowOp [(None, 50, 1, 128)] 0           activation_67[0][0]              \n",
      "                                                                 tf_op_layer_ExpandDims_3[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "lambda_7 (Lambda)               (None, 1, 1, 128)    0           tf_op_layer_mul_3[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "flatten_3 (Flatten)             (None, 128)          0           lambda_7[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "gaussian_dropout_70 (GaussianDr (None, 128)          0           flatten_3[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "dense_6 (Dense)                 (None, 128)          16512       gaussian_dropout_70[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "gaussian_dropout_71 (GaussianDr (None, 128)          0           dense_6[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "dense_7 (Dense)                 (None, 4)            516         gaussian_dropout_71[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "prediction (Lambda)             (None, 4)            0           dense_7[0][0]                    \n",
      "==================================================================================================\n",
      "Total params: 110,836\n",
      "Trainable params: 110,836\n",
      "Non-trainable params: 0\n",
      "__________________________________________________________________________________________________\n",
      "Failed to import pydot. You must install pydot and graphviz for `pydotprint` to work.\n"
     ]
    }
   ],
   "source": [
    "# Instantiate Graph Neural Network\n",
    "graph = GraphNet()\n",
    "# Overwrite default model parameters\n",
    "params = {\n",
    "    \"N_t\": N_t,\n",
    "    \"dropout_rate\": 0.15,\n",
    "    \"activation\": \"relu\",\n",
    "}\n",
    "# Set parameters\n",
    "graph.set_params(params)\n",
    "# Construct a graph model\n",
    "model = graph.construct()\n",
    "# Print out a summary\n",
    "model.summary()\n",
    "# Output model summary to a PNG image (if possible)\n",
    "plot_model(model, show_shapes=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train for 34 steps, validate for 8 steps\n",
      "Epoch 1/500\n",
      "34/34 - 9s - loss: 0.4707 - val_loss: 0.4283\n",
      "Epoch 2/500\n",
      "34/34 - 7s - loss: 0.3941 - val_loss: 0.3638\n",
      "Epoch 3/500\n",
      "34/34 - 7s - loss: 0.3569 - val_loss: 0.3420\n",
      "Epoch 4/500\n",
      "34/34 - 8s - loss: 0.3446 - val_loss: 0.3429\n",
      "Epoch 5/500\n",
      "34/34 - 8s - loss: 0.3404 - val_loss: 0.3389\n",
      "Epoch 6/500\n",
      "34/34 - 7s - loss: 0.3372 - val_loss: 0.3417\n",
      "Epoch 7/500\n",
      "34/34 - 7s - loss: 0.3353 - val_loss: 0.3373\n",
      "Epoch 8/500\n",
      "34/34 - 7s - loss: 0.3353 - val_loss: 0.3318\n",
      "Epoch 9/500\n",
      "34/34 - 7s - loss: 0.3315 - val_loss: 0.3265\n",
      "Epoch 10/500\n",
      "34/34 - 7s - loss: 0.3257 - val_loss: 0.3227\n",
      "Epoch 11/500\n",
      "34/34 - 8s - loss: 0.3191 - val_loss: 0.3204\n",
      "Epoch 12/500\n",
      "34/34 - 7s - loss: 0.3122 - val_loss: 0.3077\n",
      "Epoch 13/500\n",
      "34/34 - 7s - loss: 0.3016 - val_loss: 0.2999\n",
      "Epoch 14/500\n",
      "34/34 - 7s - loss: 0.2935 - val_loss: 0.2855\n",
      "Epoch 15/500\n",
      "34/34 - 7s - loss: 0.2881 - val_loss: 0.2771\n",
      "Epoch 16/500\n",
      "34/34 - 8s - loss: 0.2774 - val_loss: 0.2673\n",
      "Epoch 17/500\n",
      "34/34 - 7s - loss: 0.2701 - val_loss: 0.2590\n",
      "Epoch 18/500\n",
      "34/34 - 7s - loss: 0.2605 - val_loss: 0.2494\n",
      "Epoch 19/500\n",
      "34/34 - 7s - loss: 0.2539 - val_loss: 0.2422\n",
      "Epoch 20/500\n",
      "34/34 - 7s - loss: 0.2433 - val_loss: 0.2367\n",
      "Epoch 21/500\n",
      "34/34 - 7s - loss: 0.2412 - val_loss: 0.2254\n",
      "Epoch 22/500\n",
      "34/34 - 7s - loss: 0.2405 - val_loss: 0.2266\n",
      "Epoch 23/500\n",
      "34/34 - 7s - loss: 0.2335 - val_loss: 0.2178\n",
      "Epoch 24/500\n",
      "34/34 - 7s - loss: 0.2304 - val_loss: 0.2144\n",
      "Epoch 25/500\n",
      "34/34 - 7s - loss: 0.2247 - val_loss: 0.2106\n",
      "Epoch 26/500\n",
      "34/34 - 7s - loss: 0.2236 - val_loss: 0.2066\n",
      "Epoch 27/500\n",
      "34/34 - 7s - loss: 0.2204 - val_loss: 0.2033\n",
      "Epoch 28/500\n",
      "34/34 - 7s - loss: 0.2168 - val_loss: 0.2068\n",
      "Epoch 29/500\n",
      "34/34 - 8s - loss: 0.2153 - val_loss: 0.1960\n",
      "Epoch 30/500\n",
      "34/34 - 7s - loss: 0.2095 - val_loss: 0.1969\n",
      "Epoch 31/500\n",
      "34/34 - 7s - loss: 0.2135 - val_loss: 0.1985\n",
      "Epoch 32/500\n",
      "34/34 - 7s - loss: 0.2107 - val_loss: 0.1915\n",
      "Epoch 33/500\n",
      "34/34 - 7s - loss: 0.2096 - val_loss: 0.1938\n",
      "Epoch 34/500\n",
      "34/34 - 7s - loss: 0.2023 - val_loss: 0.1935\n",
      "Epoch 35/500\n",
      "34/34 - 7s - loss: 0.2030 - val_loss: 0.1892\n",
      "Epoch 36/500\n",
      "34/34 - 8s - loss: 0.1991 - val_loss: 0.1877\n",
      "Epoch 37/500\n",
      "34/34 - 8s - loss: 0.2024 - val_loss: 0.1859\n",
      "Epoch 38/500\n",
      "34/34 - 7s - loss: 0.1994 - val_loss: 0.1818\n",
      "Epoch 39/500\n",
      "34/34 - 7s - loss: 0.1967 - val_loss: 0.1847\n",
      "Epoch 40/500\n",
      "34/34 - 7s - loss: 0.1970 - val_loss: 0.1834\n",
      "Epoch 41/500\n",
      "34/34 - 7s - loss: 0.1947 - val_loss: 0.1805\n",
      "Epoch 42/500\n",
      "34/34 - 8s - loss: 0.1911 - val_loss: 0.1836\n",
      "Epoch 43/500\n",
      "34/34 - 7s - loss: 0.1935 - val_loss: 0.1823\n",
      "Epoch 44/500\n",
      "34/34 - 8s - loss: 0.1949 - val_loss: 0.1770\n",
      "Epoch 45/500\n",
      "34/34 - 7s - loss: 0.1921 - val_loss: 0.1780\n",
      "Epoch 46/500\n",
      "34/34 - 7s - loss: 0.1904 - val_loss: 0.1735\n",
      "Epoch 47/500\n",
      "34/34 - 7s - loss: 0.1907 - val_loss: 0.1807\n",
      "Epoch 48/500\n",
      "34/34 - 7s - loss: 0.1891 - val_loss: 0.1739\n",
      "Epoch 49/500\n",
      "34/34 - 7s - loss: 0.1873 - val_loss: 0.1781\n",
      "Epoch 50/500\n",
      "34/34 - 7s - loss: 0.1858 - val_loss: 0.1763\n",
      "Epoch 51/500\n",
      "34/34 - 7s - loss: 0.1887 - val_loss: 0.1754\n",
      "Epoch 52/500\n",
      "34/34 - 7s - loss: 0.1825 - val_loss: 0.1742\n",
      "Epoch 53/500\n",
      "34/34 - 7s - loss: 0.1833 - val_loss: 0.1753\n",
      "Epoch 54/500\n",
      "34/34 - 7s - loss: 0.1863 - val_loss: 0.1687\n",
      "Epoch 55/500\n",
      "34/34 - 7s - loss: 0.1834 - val_loss: 0.1709\n",
      "Epoch 56/500\n",
      "34/34 - 7s - loss: 0.1846 - val_loss: 0.1720\n",
      "Epoch 57/500\n",
      "34/34 - 7s - loss: 0.1810 - val_loss: 0.1693\n",
      "Epoch 58/500\n",
      "34/34 - 7s - loss: 0.1809 - val_loss: 0.1717\n",
      "Epoch 59/500\n",
      "34/34 - 7s - loss: 0.1806 - val_loss: 0.1705\n",
      "Epoch 60/500\n",
      "34/34 - 7s - loss: 0.1802 - val_loss: 0.1703\n",
      "Epoch 61/500\n",
      "34/34 - 7s - loss: 0.1783 - val_loss: 0.1694\n",
      "Epoch 62/500\n",
      "34/34 - 7s - loss: 0.1777 - val_loss: 0.1677\n",
      "Epoch 63/500\n",
      "34/34 - 7s - loss: 0.1788 - val_loss: 0.1671\n",
      "Epoch 64/500\n",
      "34/34 - 7s - loss: 0.1776 - val_loss: 0.1680\n",
      "Epoch 65/500\n",
      "34/34 - 7s - loss: 0.1747 - val_loss: 0.1698\n",
      "Epoch 66/500\n",
      "34/34 - 7s - loss: 0.1763 - val_loss: 0.1660\n",
      "Epoch 67/500\n",
      "34/34 - 7s - loss: 0.1740 - val_loss: 0.1661\n",
      "Epoch 68/500\n",
      "34/34 - 7s - loss: 0.1743 - val_loss: 0.1644\n",
      "Epoch 69/500\n",
      "34/34 - 7s - loss: 0.1757 - val_loss: 0.1680\n",
      "Epoch 70/500\n",
      "34/34 - 8s - loss: 0.1734 - val_loss: 0.1642\n",
      "Epoch 71/500\n",
      "34/34 - 8s - loss: 0.1712 - val_loss: 0.1632\n",
      "Epoch 72/500\n",
      "34/34 - 7s - loss: 0.1730 - val_loss: 0.1647\n",
      "Epoch 73/500\n",
      "34/34 - 8s - loss: 0.1707 - val_loss: 0.1633\n",
      "Epoch 74/500\n",
      "34/34 - 7s - loss: 0.1707 - val_loss: 0.1599\n",
      "Epoch 75/500\n",
      "34/34 - 8s - loss: 0.1711 - val_loss: 0.1567\n",
      "Epoch 76/500\n",
      "34/34 - 7s - loss: 0.1678 - val_loss: 0.1600\n",
      "Epoch 77/500\n",
      "34/34 - 7s - loss: 0.1681 - val_loss: 0.1633\n",
      "Epoch 78/500\n",
      "34/34 - 8s - loss: 0.1670 - val_loss: 0.1580\n",
      "Epoch 79/500\n",
      "34/34 - 7s - loss: 0.1646 - val_loss: 0.1601\n",
      "Epoch 80/500\n",
      "34/34 - 7s - loss: 0.1659 - val_loss: 0.1636\n",
      "Epoch 81/500\n",
      "34/34 - 7s - loss: 0.1692 - val_loss: 0.1576\n",
      "Epoch 82/500\n",
      "34/34 - 8s - loss: 0.1656 - val_loss: 0.1574\n",
      "Epoch 83/500\n",
      "34/34 - 8s - loss: 0.1643 - val_loss: 0.1550\n",
      "Epoch 84/500\n",
      "34/34 - 8s - loss: 0.1658 - val_loss: 0.1547\n",
      "Epoch 85/500\n",
      "34/34 - 8s - loss: 0.1639 - val_loss: 0.1573\n",
      "Epoch 86/500\n",
      "34/34 - 8s - loss: 0.1627 - val_loss: 0.1520\n",
      "Epoch 87/500\n",
      "34/34 - 7s - loss: 0.1642 - val_loss: 0.1538\n",
      "Epoch 88/500\n",
      "34/34 - 8s - loss: 0.1622 - val_loss: 0.1535\n",
      "Epoch 89/500\n",
      "34/34 - 7s - loss: 0.1627 - val_loss: 0.1542\n",
      "Epoch 90/500\n",
      "34/34 - 7s - loss: 0.1595 - val_loss: 0.1541\n",
      "Epoch 91/500\n",
      "34/34 - 8s - loss: 0.1612 - val_loss: 0.1519\n",
      "Epoch 92/500\n",
      "34/34 - 8s - loss: 0.1605 - val_loss: 0.1526\n",
      "Epoch 93/500\n",
      "34/34 - 8s - loss: 0.1596 - val_loss: 0.1511\n",
      "Epoch 94/500\n",
      "34/34 - 7s - loss: 0.1598 - val_loss: 0.1499\n",
      "Epoch 95/500\n",
      "34/34 - 7s - loss: 0.1604 - val_loss: 0.1512\n",
      "Epoch 96/500\n",
      "34/34 - 7s - loss: 0.1596 - val_loss: 0.1513\n",
      "Epoch 97/500\n",
      "34/34 - 7s - loss: 0.1573 - val_loss: 0.1513\n",
      "Epoch 98/500\n",
      "34/34 - 7s - loss: 0.1577 - val_loss: 0.1485\n",
      "Epoch 99/500\n",
      "34/34 - 7s - loss: 0.1576 - val_loss: 0.1490\n",
      "Epoch 100/500\n",
      "34/34 - 8s - loss: 0.1566 - val_loss: 0.1482\n",
      "Epoch 101/500\n",
      "34/34 - 8s - loss: 0.1558 - val_loss: 0.1453\n",
      "Epoch 102/500\n",
      "34/34 - 8s - loss: 0.1571 - val_loss: 0.1481\n",
      "Epoch 103/500\n",
      "34/34 - 7s - loss: 0.1545 - val_loss: 0.1515\n",
      "Epoch 104/500\n",
      "34/34 - 7s - loss: 0.1554 - val_loss: 0.1487\n",
      "Epoch 105/500\n",
      "34/34 - 7s - loss: 0.1523 - val_loss: 0.1428\n",
      "Epoch 106/500\n",
      "34/34 - 7s - loss: 0.1542 - val_loss: 0.1436\n",
      "Epoch 107/500\n",
      "34/34 - 8s - loss: 0.1542 - val_loss: 0.1436\n",
      "Epoch 108/500\n",
      "34/34 - 8s - loss: 0.1526 - val_loss: 0.1444\n",
      "Epoch 109/500\n",
      "34/34 - 8s - loss: 0.1537 - val_loss: 0.1456\n",
      "Epoch 110/500\n",
      "34/34 - 7s - loss: 0.1518 - val_loss: 0.1456\n",
      "Epoch 111/500\n",
      "34/34 - 8s - loss: 0.1524 - val_loss: 0.1438\n",
      "Epoch 112/500\n",
      "34/34 - 7s - loss: 0.1513 - val_loss: 0.1437\n",
      "Epoch 113/500\n",
      "34/34 - 8s - loss: 0.1502 - val_loss: 0.1435\n",
      "Epoch 114/500\n",
      "34/34 - 7s - loss: 0.1506 - val_loss: 0.1445\n",
      "Epoch 115/500\n",
      "34/34 - 8s - loss: 0.1490 - val_loss: 0.1427\n",
      "Epoch 116/500\n",
      "34/34 - 8s - loss: 0.1500 - val_loss: 0.1400\n",
      "Epoch 117/500\n",
      "34/34 - 8s - loss: 0.1503 - val_loss: 0.1456\n",
      "Epoch 118/500\n",
      "34/34 - 7s - loss: 0.1494 - val_loss: 0.1416\n",
      "Epoch 119/500\n",
      "34/34 - 7s - loss: 0.1491 - val_loss: 0.1401\n",
      "Epoch 120/500\n",
      "34/34 - 7s - loss: 0.1478 - val_loss: 0.1407\n",
      "Epoch 121/500\n",
      "34/34 - 8s - loss: 0.1481 - val_loss: 0.1382\n",
      "Epoch 122/500\n",
      "34/34 - 8s - loss: 0.1492 - val_loss: 0.1412\n",
      "Epoch 123/500\n",
      "34/34 - 7s - loss: 0.1470 - val_loss: 0.1420\n",
      "Epoch 124/500\n",
      "34/34 - 7s - loss: 0.1473 - val_loss: 0.1386\n",
      "Epoch 125/500\n",
      "34/34 - 7s - loss: 0.1482 - val_loss: 0.1406\n",
      "Epoch 126/500\n",
      "34/34 - 8s - loss: 0.1447 - val_loss: 0.1371\n",
      "Epoch 127/500\n",
      "34/34 - 8s - loss: 0.1466 - val_loss: 0.1401\n",
      "Epoch 128/500\n",
      "34/34 - 7s - loss: 0.1465 - val_loss: 0.1402\n",
      "Epoch 129/500\n",
      "34/34 - 7s - loss: 0.1441 - val_loss: 0.1412\n",
      "Epoch 130/500\n",
      "34/34 - 7s - loss: 0.1438 - val_loss: 0.1387\n",
      "Epoch 131/500\n",
      "34/34 - 7s - loss: 0.1449 - val_loss: 0.1418\n",
      "Epoch 132/500\n",
      "34/34 - 7s - loss: 0.1440 - val_loss: 0.1393\n",
      "Epoch 133/500\n",
      "34/34 - 7s - loss: 0.1420 - val_loss: 0.1378\n",
      "Epoch 134/500\n",
      "34/34 - 7s - loss: 0.1437 - val_loss: 0.1364\n",
      "Epoch 135/500\n",
      "34/34 - 8s - loss: 0.1448 - val_loss: 0.1347\n",
      "Epoch 136/500\n",
      "34/34 - 8s - loss: 0.1430 - val_loss: 0.1341\n",
      "Epoch 137/500\n",
      "34/34 - 7s - loss: 0.1420 - val_loss: 0.1364\n",
      "Epoch 138/500\n",
      "34/34 - 7s - loss: 0.1434 - val_loss: 0.1358\n",
      "Epoch 139/500\n",
      "34/34 - 8s - loss: 0.1449 - val_loss: 0.1375\n",
      "Epoch 140/500\n",
      "34/34 - 7s - loss: 0.1424 - val_loss: 0.1366\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 141/500\n",
      "34/34 - 8s - loss: 0.1428 - val_loss: 0.1383\n",
      "Epoch 142/500\n",
      "34/34 - 7s - loss: 0.1408 - val_loss: 0.1358\n",
      "Epoch 143/500\n",
      "34/34 - 7s - loss: 0.1398 - val_loss: 0.1347\n",
      "Epoch 144/500\n",
      "34/34 - 7s - loss: 0.1434 - val_loss: 0.1363\n",
      "Epoch 145/500\n",
      "34/34 - 7s - loss: 0.1395 - val_loss: 0.1337\n",
      "Epoch 146/500\n",
      "34/34 - 7s - loss: 0.1394 - val_loss: 0.1333\n",
      "Epoch 147/500\n",
      "34/34 - 8s - loss: 0.1406 - val_loss: 0.1358\n",
      "Epoch 148/500\n",
      "34/34 - 7s - loss: 0.1410 - val_loss: 0.1350\n",
      "Epoch 149/500\n",
      "34/34 - 7s - loss: 0.1404 - val_loss: 0.1340\n",
      "Epoch 150/500\n",
      "34/34 - 8s - loss: 0.1372 - val_loss: 0.1327\n",
      "Epoch 151/500\n",
      "34/34 - 8s - loss: 0.1387 - val_loss: 0.1338\n",
      "Epoch 152/500\n",
      "34/34 - 7s - loss: 0.1388 - val_loss: 0.1334\n",
      "Epoch 153/500\n",
      "34/34 - 8s - loss: 0.1383 - val_loss: 0.1350\n",
      "Epoch 154/500\n",
      "34/34 - 7s - loss: 0.1365 - val_loss: 0.1378\n",
      "Epoch 155/500\n",
      "34/34 - 8s - loss: 0.1392 - val_loss: 0.1303\n",
      "Epoch 156/500\n",
      "34/34 - 7s - loss: 0.1359 - val_loss: 0.1344\n",
      "Epoch 157/500\n",
      "34/34 - 7s - loss: 0.1375 - val_loss: 0.1336\n",
      "Epoch 158/500\n",
      "34/34 - 7s - loss: 0.1383 - val_loss: 0.1327\n",
      "Epoch 159/500\n",
      "34/34 - 7s - loss: 0.1369 - val_loss: 0.1308\n",
      "Epoch 160/500\n",
      "34/34 - 8s - loss: 0.1376 - val_loss: 0.1308\n",
      "Epoch 161/500\n",
      "34/34 - 7s - loss: 0.1370 - val_loss: 0.1333\n",
      "Epoch 162/500\n",
      "34/34 - 8s - loss: 0.1386 - val_loss: 0.1309\n",
      "Epoch 163/500\n",
      "34/34 - 8s - loss: 0.1356 - val_loss: 0.1276\n",
      "Epoch 164/500\n",
      "34/34 - 8s - loss: 0.1366 - val_loss: 0.1328\n",
      "Epoch 165/500\n",
      "34/34 - 7s - loss: 0.1362 - val_loss: 0.1317\n",
      "Epoch 166/500\n",
      "34/34 - 7s - loss: 0.1368 - val_loss: 0.1297\n",
      "Epoch 167/500\n",
      "34/34 - 7s - loss: 0.1344 - val_loss: 0.1336\n",
      "Epoch 168/500\n",
      "34/34 - 7s - loss: 0.1380 - val_loss: 0.1331\n",
      "Epoch 169/500\n",
      "34/34 - 8s - loss: 0.1361 - val_loss: 0.1312\n",
      "Epoch 170/500\n",
      "34/34 - 7s - loss: 0.1340 - val_loss: 0.1308\n",
      "Epoch 171/500\n",
      "34/34 - 7s - loss: 0.1336 - val_loss: 0.1305\n",
      "Epoch 172/500\n",
      "34/34 - 7s - loss: 0.1365 - val_loss: 0.1290\n",
      "Epoch 173/500\n",
      "34/34 - 7s - loss: 0.1345 - val_loss: 0.1280\n",
      "Epoch 174/500\n",
      "34/34 - 8s - loss: 0.1356 - val_loss: 0.1294\n",
      "Epoch 175/500\n",
      "34/34 - 8s - loss: 0.1352 - val_loss: 0.1283\n",
      "Epoch 176/500\n",
      "34/34 - 7s - loss: 0.1346 - val_loss: 0.1318\n",
      "Epoch 177/500\n",
      "34/34 - 7s - loss: 0.1327 - val_loss: 0.1310\n",
      "Epoch 178/500\n",
      "34/34 - 7s - loss: 0.1341 - val_loss: 0.1300\n",
      "Epoch 179/500\n",
      "34/34 - 7s - loss: 0.1347 - val_loss: 0.1310\n",
      "Epoch 180/500\n",
      "34/34 - 7s - loss: 0.1335 - val_loss: 0.1306\n",
      "Epoch 181/500\n",
      "34/34 - 7s - loss: 0.1339 - val_loss: 0.1311\n",
      "Epoch 182/500\n",
      "34/34 - 7s - loss: 0.1338 - val_loss: 0.1294\n",
      "Epoch 183/500\n",
      "34/34 - 7s - loss: 0.1328 - val_loss: 0.1293\n",
      "Epoch 184/500\n",
      "34/34 - 7s - loss: 0.1339 - val_loss: 0.1287\n",
      "Epoch 185/500\n",
      "34/34 - 8s - loss: 0.1332 - val_loss: 0.1278\n",
      "Epoch 186/500\n",
      "34/34 - 7s - loss: 0.1312 - val_loss: 0.1289\n",
      "Epoch 187/500\n",
      "34/34 - 7s - loss: 0.1319 - val_loss: 0.1284\n",
      "Epoch 188/500\n",
      "34/34 - 7s - loss: 0.1337 - val_loss: 0.1256\n",
      "Epoch 189/500\n",
      "34/34 - 7s - loss: 0.1333 - val_loss: 0.1269\n",
      "Epoch 190/500\n",
      "34/34 - 7s - loss: 0.1319 - val_loss: 0.1263\n",
      "Epoch 191/500\n",
      "34/34 - 8s - loss: 0.1315 - val_loss: 0.1278\n",
      "Epoch 192/500\n",
      "34/34 - 7s - loss: 0.1319 - val_loss: 0.1301\n",
      "Epoch 193/500\n",
      "34/34 - 7s - loss: 0.1333 - val_loss: 0.1307\n",
      "Epoch 194/500\n",
      "34/34 - 7s - loss: 0.1324 - val_loss: 0.1277\n",
      "Epoch 195/500\n",
      "34/34 - 7s - loss: 0.1309 - val_loss: 0.1294\n",
      "Epoch 196/500\n",
      "34/34 - 7s - loss: 0.1307 - val_loss: 0.1261\n",
      "Epoch 197/500\n",
      "34/34 - 7s - loss: 0.1307 - val_loss: 0.1286\n",
      "Epoch 198/500\n",
      "34/34 - 7s - loss: 0.1294 - val_loss: 0.1263\n",
      "Epoch 199/500\n",
      "34/34 - 7s - loss: 0.1321 - val_loss: 0.1290\n",
      "Epoch 200/500\n",
      "34/34 - 8s - loss: 0.1306 - val_loss: 0.1264\n",
      "Epoch 201/500\n",
      "34/34 - 8s - loss: 0.1309 - val_loss: 0.1258\n",
      "Epoch 202/500\n",
      "34/34 - 7s - loss: 0.1312 - val_loss: 0.1271\n",
      "Epoch 203/500\n",
      "34/34 - 7s - loss: 0.1302 - val_loss: 0.1250\n",
      "Epoch 204/500\n",
      "34/34 - 7s - loss: 0.1299 - val_loss: 0.1273\n",
      "Epoch 205/500\n",
      "34/34 - 7s - loss: 0.1286 - val_loss: 0.1244\n",
      "Epoch 206/500\n",
      "34/34 - 7s - loss: 0.1292 - val_loss: 0.1260\n",
      "Epoch 207/500\n",
      "34/34 - 8s - loss: 0.1303 - val_loss: 0.1216\n",
      "Epoch 208/500\n",
      "34/34 - 7s - loss: 0.1305 - val_loss: 0.1261\n",
      "Epoch 209/500\n",
      "34/34 - 7s - loss: 0.1297 - val_loss: 0.1275\n",
      "Epoch 210/500\n",
      "34/34 - 8s - loss: 0.1296 - val_loss: 0.1243\n",
      "Epoch 211/500\n",
      "34/34 - 7s - loss: 0.1274 - val_loss: 0.1257\n",
      "Epoch 212/500\n",
      "34/34 - 7s - loss: 0.1318 - val_loss: 0.1246\n",
      "Epoch 213/500\n",
      "34/34 - 8s - loss: 0.1296 - val_loss: 0.1267\n",
      "Epoch 214/500\n",
      "34/34 - 8s - loss: 0.1307 - val_loss: 0.1258\n",
      "Epoch 215/500\n",
      "34/34 - 8s - loss: 0.1290 - val_loss: 0.1256\n",
      "Epoch 216/500\n",
      "34/34 - 8s - loss: 0.1289 - val_loss: 0.1262\n",
      "Epoch 217/500\n",
      "34/34 - 8s - loss: 0.1301 - val_loss: 0.1235\n",
      "Epoch 218/500\n",
      "34/34 - 8s - loss: 0.1291 - val_loss: 0.1252\n",
      "Epoch 219/500\n",
      "34/34 - 8s - loss: 0.1304 - val_loss: 0.1282\n",
      "Epoch 220/500\n",
      "34/34 - 7s - loss: 0.1286 - val_loss: 0.1257\n",
      "Epoch 221/500\n",
      "34/34 - 8s - loss: 0.1294 - val_loss: 0.1234\n",
      "Epoch 222/500\n",
      "34/34 - 7s - loss: 0.1280 - val_loss: 0.1240\n",
      "Epoch 223/500\n",
      "34/34 - 8s - loss: 0.1277 - val_loss: 0.1245\n",
      "Epoch 224/500\n",
      "34/34 - 8s - loss: 0.1256 - val_loss: 0.1215\n",
      "Epoch 225/500\n",
      "34/34 - 8s - loss: 0.1282 - val_loss: 0.1234\n",
      "Epoch 226/500\n",
      "34/34 - 7s - loss: 0.1278 - val_loss: 0.1258\n",
      "Epoch 227/500\n",
      "34/34 - 7s - loss: 0.1279 - val_loss: 0.1264\n",
      "Epoch 228/500\n",
      "34/34 - 8s - loss: 0.1276 - val_loss: 0.1238\n",
      "Epoch 229/500\n",
      "34/34 - 7s - loss: 0.1266 - val_loss: 0.1237\n",
      "Epoch 230/500\n",
      "34/34 - 7s - loss: 0.1267 - val_loss: 0.1224\n",
      "Epoch 231/500\n",
      "34/34 - 7s - loss: 0.1279 - val_loss: 0.1253\n",
      "Epoch 232/500\n",
      "34/34 - 7s - loss: 0.1276 - val_loss: 0.1236\n",
      "Epoch 233/500\n",
      "34/34 - 7s - loss: 0.1273 - val_loss: 0.1242\n",
      "Epoch 234/500\n",
      "34/34 - 8s - loss: 0.1280 - val_loss: 0.1247\n",
      "Epoch 235/500\n",
      "34/34 - 7s - loss: 0.1266 - val_loss: 0.1224\n",
      "Epoch 236/500\n",
      "34/34 - 7s - loss: 0.1280 - val_loss: 0.1233\n",
      "Epoch 237/500\n",
      "34/34 - 7s - loss: 0.1278 - val_loss: 0.1244\n",
      "Epoch 238/500\n",
      "34/34 - 8s - loss: 0.1275 - val_loss: 0.1247\n",
      "Epoch 239/500\n",
      "34/34 - 7s - loss: 0.1278 - val_loss: 0.1241\n",
      "Epoch 240/500\n",
      "34/34 - 7s - loss: 0.1267 - val_loss: 0.1252\n",
      "Epoch 241/500\n",
      "34/34 - 7s - loss: 0.1262 - val_loss: 0.1257\n",
      "Epoch 242/500\n",
      "34/34 - 7s - loss: 0.1266 - val_loss: 0.1248\n",
      "Epoch 243/500\n",
      "34/34 - 7s - loss: 0.1259 - val_loss: 0.1220\n",
      "Epoch 244/500\n",
      "34/34 - 7s - loss: 0.1267 - val_loss: 0.1241\n",
      "Epoch 245/500\n",
      "34/34 - 7s - loss: 0.1242 - val_loss: 0.1226\n",
      "Epoch 246/500\n",
      "34/34 - 7s - loss: 0.1263 - val_loss: 0.1209\n",
      "Epoch 247/500\n",
      "34/34 - 7s - loss: 0.1245 - val_loss: 0.1240\n",
      "Epoch 248/500\n",
      "34/34 - 8s - loss: 0.1257 - val_loss: 0.1230\n",
      "Epoch 249/500\n",
      "34/34 - 7s - loss: 0.1261 - val_loss: 0.1214\n",
      "Epoch 250/500\n",
      "34/34 - 7s - loss: 0.1259 - val_loss: 0.1225\n",
      "Epoch 251/500\n",
      "34/34 - 7s - loss: 0.1254 - val_loss: 0.1211\n",
      "Epoch 252/500\n",
      "34/34 - 7s - loss: 0.1251 - val_loss: 0.1228\n",
      "Epoch 253/500\n",
      "34/34 - 7s - loss: 0.1251 - val_loss: 0.1203\n",
      "Epoch 254/500\n",
      "34/34 - 7s - loss: 0.1258 - val_loss: 0.1247\n",
      "Epoch 255/500\n",
      "34/34 - 7s - loss: 0.1264 - val_loss: 0.1242\n",
      "Epoch 256/500\n",
      "34/34 - 7s - loss: 0.1269 - val_loss: 0.1224\n",
      "Epoch 257/500\n",
      "34/34 - 7s - loss: 0.1242 - val_loss: 0.1244\n",
      "Epoch 258/500\n",
      "34/34 - 7s - loss: 0.1247 - val_loss: 0.1209\n",
      "Epoch 259/500\n",
      "34/34 - 7s - loss: 0.1247 - val_loss: 0.1249\n",
      "Epoch 260/500\n",
      "34/34 - 7s - loss: 0.1249 - val_loss: 0.1232\n",
      "Epoch 261/500\n",
      "34/34 - 8s - loss: 0.1260 - val_loss: 0.1232\n",
      "Epoch 262/500\n",
      "34/34 - 7s - loss: 0.1248 - val_loss: 0.1232\n",
      "Epoch 263/500\n",
      "34/34 - 7s - loss: 0.1250 - val_loss: 0.1198\n",
      "Epoch 264/500\n",
      "34/34 - 8s - loss: 0.1253 - val_loss: 0.1224\n",
      "Epoch 265/500\n",
      "34/34 - 7s - loss: 0.1255 - val_loss: 0.1214\n",
      "Epoch 266/500\n",
      "34/34 - 7s - loss: 0.1258 - val_loss: 0.1221\n",
      "Epoch 267/500\n",
      "34/34 - 7s - loss: 0.1248 - val_loss: 0.1235\n",
      "Epoch 268/500\n",
      "34/34 - 7s - loss: 0.1239 - val_loss: 0.1232\n",
      "Epoch 269/500\n",
      "34/34 - 8s - loss: 0.1259 - val_loss: 0.1220\n",
      "Epoch 270/500\n",
      "34/34 - 7s - loss: 0.1243 - val_loss: 0.1206\n",
      "Epoch 271/500\n",
      "34/34 - 7s - loss: 0.1236 - val_loss: 0.1212\n",
      "Epoch 272/500\n",
      "34/34 - 7s - loss: 0.1253 - val_loss: 0.1220\n",
      "Epoch 273/500\n",
      "34/34 - 7s - loss: 0.1224 - val_loss: 0.1229\n",
      "Epoch 274/500\n",
      "34/34 - 7s - loss: 0.1225 - val_loss: 0.1212\n",
      "Epoch 275/500\n",
      "34/34 - 7s - loss: 0.1237 - val_loss: 0.1197\n",
      "Epoch 276/500\n",
      "34/34 - 7s - loss: 0.1231 - val_loss: 0.1211\n",
      "Epoch 277/500\n",
      "34/34 - 8s - loss: 0.1232 - val_loss: 0.1219\n",
      "Epoch 278/500\n",
      "34/34 - 8s - loss: 0.1234 - val_loss: 0.1194\n",
      "Epoch 279/500\n",
      "34/34 - 8s - loss: 0.1246 - val_loss: 0.1213\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 280/500\n",
      "34/34 - 7s - loss: 0.1229 - val_loss: 0.1222\n",
      "Epoch 281/500\n",
      "34/34 - 7s - loss: 0.1239 - val_loss: 0.1207\n",
      "Epoch 282/500\n",
      "34/34 - 7s - loss: 0.1209 - val_loss: 0.1229\n",
      "Epoch 283/500\n",
      "34/34 - 7s - loss: 0.1231 - val_loss: 0.1201\n",
      "Epoch 284/500\n",
      "34/34 - 7s - loss: 0.1225 - val_loss: 0.1225\n",
      "Epoch 285/500\n",
      "34/34 - 8s - loss: 0.1245 - val_loss: 0.1189\n",
      "Epoch 286/500\n",
      "34/34 - 7s - loss: 0.1229 - val_loss: 0.1219\n",
      "Epoch 287/500\n",
      "34/34 - 7s - loss: 0.1216 - val_loss: 0.1181\n",
      "Epoch 288/500\n",
      "34/34 - 7s - loss: 0.1225 - val_loss: 0.1202\n",
      "Epoch 289/500\n",
      "34/34 - 7s - loss: 0.1235 - val_loss: 0.1216\n",
      "Epoch 290/500\n",
      "34/34 - 7s - loss: 0.1212 - val_loss: 0.1209\n",
      "Epoch 291/500\n",
      "34/34 - 7s - loss: 0.1204 - val_loss: 0.1204\n",
      "Epoch 292/500\n",
      "34/34 - 7s - loss: 0.1244 - val_loss: 0.1209\n",
      "Epoch 293/500\n",
      "34/34 - 7s - loss: 0.1230 - val_loss: 0.1210\n",
      "Epoch 294/500\n",
      "34/34 - 8s - loss: 0.1226 - val_loss: 0.1216\n",
      "Epoch 295/500\n",
      "34/34 - 7s - loss: 0.1208 - val_loss: 0.1205\n",
      "Epoch 296/500\n",
      "34/34 - 7s - loss: 0.1214 - val_loss: 0.1224\n",
      "Epoch 297/500\n",
      "34/34 - 7s - loss: 0.1209 - val_loss: 0.1194\n",
      "Epoch 298/500\n",
      "34/34 - 7s - loss: 0.1212 - val_loss: 0.1214\n",
      "Epoch 299/500\n",
      "34/34 - 7s - loss: 0.1219 - val_loss: 0.1215\n",
      "Epoch 300/500\n",
      "34/34 - 7s - loss: 0.1217 - val_loss: 0.1186\n",
      "Epoch 301/500\n",
      "34/34 - 7s - loss: 0.1225 - val_loss: 0.1212\n",
      "Epoch 302/500\n",
      "34/34 - 7s - loss: 0.1209 - val_loss: 0.1214\n",
      "Epoch 303/500\n",
      "34/34 - 7s - loss: 0.1203 - val_loss: 0.1215\n",
      "Epoch 304/500\n",
      "34/34 - 7s - loss: 0.1216 - val_loss: 0.1214\n",
      "Epoch 305/500\n",
      "34/34 - 7s - loss: 0.1213 - val_loss: 0.1206\n",
      "Epoch 306/500\n",
      "34/34 - 7s - loss: 0.1209 - val_loss: 0.1202\n",
      "Epoch 307/500\n",
      "34/34 - 7s - loss: 0.1229 - val_loss: 0.1216\n",
      "Epoch 308/500\n",
      "34/34 - 7s - loss: 0.1205 - val_loss: 0.1222\n",
      "Epoch 309/500\n",
      "34/34 - 7s - loss: 0.1200 - val_loss: 0.1208\n",
      "Epoch 310/500\n",
      "34/34 - 7s - loss: 0.1217 - val_loss: 0.1198\n",
      "Epoch 311/500\n",
      "34/34 - 8s - loss: 0.1209 - val_loss: 0.1204\n",
      "Epoch 312/500\n",
      "34/34 - 8s - loss: 0.1217 - val_loss: 0.1185\n",
      "Epoch 313/500\n",
      "34/34 - 7s - loss: 0.1207 - val_loss: 0.1181\n",
      "Epoch 314/500\n",
      "34/34 - 7s - loss: 0.1191 - val_loss: 0.1189\n",
      "Epoch 315/500\n",
      "34/34 - 7s - loss: 0.1210 - val_loss: 0.1207\n",
      "Epoch 316/500\n",
      "34/34 - 7s - loss: 0.1229 - val_loss: 0.1160\n",
      "Epoch 317/500\n",
      "34/34 - 7s - loss: 0.1215 - val_loss: 0.1201\n",
      "Epoch 318/500\n",
      "34/34 - 7s - loss: 0.1208 - val_loss: 0.1198\n",
      "Epoch 319/500\n",
      "34/34 - 7s - loss: 0.1189 - val_loss: 0.1182\n",
      "Epoch 320/500\n",
      "34/34 - 7s - loss: 0.1213 - val_loss: 0.1194\n",
      "Epoch 321/500\n",
      "34/34 - 7s - loss: 0.1203 - val_loss: 0.1189\n",
      "Epoch 322/500\n",
      "34/34 - 7s - loss: 0.1199 - val_loss: 0.1212\n",
      "Epoch 323/500\n",
      "34/34 - 7s - loss: 0.1217 - val_loss: 0.1203\n",
      "Epoch 324/500\n",
      "34/34 - 7s - loss: 0.1209 - val_loss: 0.1191\n",
      "Epoch 325/500\n",
      "34/34 - 7s - loss: 0.1199 - val_loss: 0.1190\n",
      "Epoch 326/500\n",
      "34/34 - 7s - loss: 0.1191 - val_loss: 0.1198\n",
      "Epoch 327/500\n",
      "34/34 - 8s - loss: 0.1205 - val_loss: 0.1177\n",
      "Epoch 328/500\n",
      "34/34 - 7s - loss: 0.1201 - val_loss: 0.1179\n",
      "Epoch 329/500\n",
      "34/34 - 7s - loss: 0.1192 - val_loss: 0.1219\n",
      "Epoch 330/500\n",
      "34/34 - 7s - loss: 0.1198 - val_loss: 0.1184\n",
      "Epoch 331/500\n",
      "34/34 - 7s - loss: 0.1187 - val_loss: 0.1191\n",
      "Epoch 332/500\n",
      "34/34 - 7s - loss: 0.1196 - val_loss: 0.1186\n",
      "Epoch 333/500\n",
      "34/34 - 7s - loss: 0.1206 - val_loss: 0.1170\n",
      "Epoch 334/500\n",
      "34/34 - 7s - loss: 0.1182 - val_loss: 0.1190\n",
      "Epoch 335/500\n",
      "34/34 - 7s - loss: 0.1192 - val_loss: 0.1243\n",
      "Epoch 336/500\n",
      "34/34 - 7s - loss: 0.1176 - val_loss: 0.1185\n",
      "Epoch 337/500\n",
      "34/34 - 7s - loss: 0.1191 - val_loss: 0.1173\n",
      "Epoch 338/500\n",
      "34/34 - 7s - loss: 0.1195 - val_loss: 0.1170\n",
      "Epoch 339/500\n",
      "34/34 - 7s - loss: 0.1194 - val_loss: 0.1192\n",
      "Epoch 340/500\n",
      "34/34 - 7s - loss: 0.1171 - val_loss: 0.1200\n",
      "Epoch 341/500\n",
      "34/34 - 7s - loss: 0.1184 - val_loss: 0.1189\n",
      "Epoch 342/500\n",
      "34/34 - 7s - loss: 0.1185 - val_loss: 0.1184\n",
      "Epoch 343/500\n",
      "34/34 - 7s - loss: 0.1197 - val_loss: 0.1191\n",
      "Epoch 344/500\n",
      "34/34 - 7s - loss: 0.1177 - val_loss: 0.1186\n",
      "Epoch 345/500\n",
      "34/34 - 8s - loss: 0.1200 - val_loss: 0.1156\n",
      "Epoch 346/500\n",
      "34/34 - 7s - loss: 0.1178 - val_loss: 0.1177\n",
      "Epoch 347/500\n",
      "34/34 - 7s - loss: 0.1171 - val_loss: 0.1162\n",
      "Epoch 348/500\n",
      "34/34 - 7s - loss: 0.1180 - val_loss: 0.1161\n",
      "Epoch 349/500\n",
      "34/34 - 7s - loss: 0.1189 - val_loss: 0.1178\n",
      "Epoch 350/500\n",
      "34/34 - 7s - loss: 0.1182 - val_loss: 0.1154\n",
      "Epoch 351/500\n",
      "34/34 - 7s - loss: 0.1189 - val_loss: 0.1183\n",
      "Epoch 352/500\n",
      "34/34 - 7s - loss: 0.1175 - val_loss: 0.1190\n",
      "Epoch 353/500\n",
      "34/34 - 7s - loss: 0.1196 - val_loss: 0.1185\n",
      "Epoch 354/500\n",
      "34/34 - 7s - loss: 0.1186 - val_loss: 0.1174\n",
      "Epoch 355/500\n",
      "34/34 - 7s - loss: 0.1178 - val_loss: 0.1209\n",
      "Epoch 356/500\n",
      "34/34 - 7s - loss: 0.1182 - val_loss: 0.1191\n",
      "Epoch 357/500\n",
      "34/34 - 7s - loss: 0.1171 - val_loss: 0.1182\n",
      "Epoch 358/500\n",
      "34/34 - 7s - loss: 0.1176 - val_loss: 0.1158\n",
      "Epoch 359/500\n",
      "34/34 - 7s - loss: 0.1187 - val_loss: 0.1176\n",
      "Epoch 360/500\n",
      "34/34 - 7s - loss: 0.1184 - val_loss: 0.1159\n",
      "Epoch 361/500\n",
      "34/34 - 7s - loss: 0.1181 - val_loss: 0.1167\n",
      "Epoch 362/500\n",
      "34/34 - 7s - loss: 0.1184 - val_loss: 0.1188\n",
      "Epoch 363/500\n",
      "34/34 - 7s - loss: 0.1175 - val_loss: 0.1177\n",
      "Epoch 364/500\n",
      "34/34 - 7s - loss: 0.1191 - val_loss: 0.1137\n",
      "Epoch 365/500\n",
      "34/34 - 8s - loss: 0.1168 - val_loss: 0.1171\n",
      "Epoch 366/500\n",
      "34/34 - 7s - loss: 0.1175 - val_loss: 0.1163\n",
      "Epoch 367/500\n",
      "34/34 - 7s - loss: 0.1184 - val_loss: 0.1160\n",
      "Epoch 368/500\n",
      "34/34 - 8s - loss: 0.1178 - val_loss: 0.1178\n",
      "Epoch 369/500\n",
      "34/34 - 7s - loss: 0.1173 - val_loss: 0.1162\n",
      "Epoch 370/500\n",
      "34/34 - 7s - loss: 0.1181 - val_loss: 0.1179\n",
      "Epoch 371/500\n",
      "34/34 - 7s - loss: 0.1173 - val_loss: 0.1174\n",
      "Epoch 372/500\n",
      "34/34 - 7s - loss: 0.1185 - val_loss: 0.1162\n",
      "Epoch 373/500\n",
      "34/34 - 7s - loss: 0.1178 - val_loss: 0.1175\n",
      "Epoch 374/500\n",
      "34/34 - 7s - loss: 0.1162 - val_loss: 0.1168\n",
      "Epoch 375/500\n",
      "34/34 - 7s - loss: 0.1160 - val_loss: 0.1174\n",
      "Epoch 376/500\n",
      "34/34 - 8s - loss: 0.1170 - val_loss: 0.1163\n",
      "Epoch 377/500\n",
      "34/34 - 7s - loss: 0.1173 - val_loss: 0.1183\n",
      "Epoch 378/500\n",
      "34/34 - 7s - loss: 0.1162 - val_loss: 0.1160\n",
      "Epoch 379/500\n",
      "34/34 - 7s - loss: 0.1172 - val_loss: 0.1173\n",
      "Epoch 380/500\n",
      "34/34 - 7s - loss: 0.1179 - val_loss: 0.1161\n",
      "Epoch 381/500\n",
      "34/34 - 7s - loss: 0.1180 - val_loss: 0.1154\n",
      "Epoch 382/500\n",
      "34/34 - 7s - loss: 0.1170 - val_loss: 0.1165\n",
      "Epoch 383/500\n",
      "34/34 - 7s - loss: 0.1167 - val_loss: 0.1156\n",
      "Epoch 384/500\n",
      "34/34 - 7s - loss: 0.1180 - val_loss: 0.1167\n",
      "Epoch 385/500\n",
      "34/34 - 7s - loss: 0.1199 - val_loss: 0.1171\n",
      "Epoch 386/500\n",
      "34/34 - 7s - loss: 0.1163 - val_loss: 0.1173\n",
      "Epoch 387/500\n",
      "34/34 - 7s - loss: 0.1163 - val_loss: 0.1171\n",
      "Epoch 388/500\n",
      "34/34 - 7s - loss: 0.1176 - val_loss: 0.1158\n",
      "Epoch 389/500\n",
      "34/34 - 7s - loss: 0.1170 - val_loss: 0.1164\n",
      "Epoch 390/500\n",
      "34/34 - 8s - loss: 0.1159 - val_loss: 0.1146\n",
      "Epoch 391/500\n",
      "34/34 - 7s - loss: 0.1168 - val_loss: 0.1167\n",
      "Epoch 392/500\n",
      "34/34 - 7s - loss: 0.1154 - val_loss: 0.1178\n",
      "Epoch 393/500\n",
      "34/34 - 8s - loss: 0.1159 - val_loss: 0.1152\n",
      "Epoch 394/500\n",
      "34/34 - 8s - loss: 0.1141 - val_loss: 0.1152\n",
      "Epoch 395/500\n",
      "34/34 - 8s - loss: 0.1146 - val_loss: 0.1117\n",
      "Epoch 396/500\n",
      "34/34 - 8s - loss: 0.1177 - val_loss: 0.1184\n",
      "Epoch 397/500\n",
      "34/34 - 7s - loss: 0.1155 - val_loss: 0.1173\n",
      "Epoch 398/500\n",
      "34/34 - 7s - loss: 0.1153 - val_loss: 0.1166\n",
      "Epoch 399/500\n",
      "34/34 - 7s - loss: 0.1160 - val_loss: 0.1132\n",
      "Epoch 400/500\n",
      "34/34 - 7s - loss: 0.1151 - val_loss: 0.1136\n",
      "Epoch 401/500\n",
      "34/34 - 7s - loss: 0.1157 - val_loss: 0.1159\n",
      "Epoch 402/500\n",
      "34/34 - 7s - loss: 0.1161 - val_loss: 0.1152\n",
      "Epoch 403/500\n",
      "34/34 - 7s - loss: 0.1163 - val_loss: 0.1143\n",
      "Epoch 404/500\n",
      "34/34 - 7s - loss: 0.1137 - val_loss: 0.1145\n",
      "Epoch 405/500\n",
      "34/34 - 8s - loss: 0.1157 - val_loss: 0.1148\n",
      "Epoch 406/500\n",
      "34/34 - 7s - loss: 0.1152 - val_loss: 0.1168\n",
      "Epoch 407/500\n",
      "34/34 - 7s - loss: 0.1152 - val_loss: 0.1177\n",
      "Epoch 408/500\n",
      "34/34 - 7s - loss: 0.1154 - val_loss: 0.1142\n",
      "Epoch 409/500\n",
      "34/34 - 7s - loss: 0.1163 - val_loss: 0.1141\n",
      "Epoch 410/500\n",
      "34/34 - 7s - loss: 0.1154 - val_loss: 0.1151\n",
      "Epoch 411/500\n",
      "34/34 - 7s - loss: 0.1151 - val_loss: 0.1158\n",
      "Epoch 412/500\n",
      "34/34 - 7s - loss: 0.1163 - val_loss: 0.1152\n",
      "Epoch 413/500\n",
      "34/34 - 7s - loss: 0.1164 - val_loss: 0.1142\n",
      "Epoch 414/500\n",
      "34/34 - 7s - loss: 0.1158 - val_loss: 0.1166\n",
      "Epoch 415/500\n",
      "34/34 - 7s - loss: 0.1159 - val_loss: 0.1145\n",
      "Epoch 416/500\n",
      "34/34 - 7s - loss: 0.1160 - val_loss: 0.1159\n",
      "Epoch 417/500\n",
      "34/34 - 7s - loss: 0.1150 - val_loss: 0.1146\n",
      "Epoch 418/500\n",
      "34/34 - 7s - loss: 0.1157 - val_loss: 0.1141\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 419/500\n",
      "34/34 - 7s - loss: 0.1159 - val_loss: 0.1152\n",
      "Epoch 420/500\n",
      "34/34 - 8s - loss: 0.1154 - val_loss: 0.1148\n",
      "Epoch 421/500\n",
      "34/34 - 7s - loss: 0.1150 - val_loss: 0.1158\n",
      "Epoch 422/500\n",
      "34/34 - 7s - loss: 0.1153 - val_loss: 0.1150\n",
      "Epoch 423/500\n",
      "34/34 - 7s - loss: 0.1149 - val_loss: 0.1172\n",
      "Epoch 424/500\n",
      "34/34 - 7s - loss: 0.1128 - val_loss: 0.1143\n",
      "Epoch 425/500\n",
      "34/34 - 7s - loss: 0.1149 - val_loss: 0.1134\n",
      "Epoch 426/500\n",
      "34/34 - 7s - loss: 0.1162 - val_loss: 0.1130\n",
      "Epoch 427/500\n",
      "34/34 - 7s - loss: 0.1151 - val_loss: 0.1161\n",
      "Epoch 428/500\n",
      "34/34 - 7s - loss: 0.1147 - val_loss: 0.1153\n",
      "Epoch 429/500\n",
      "34/34 - 7s - loss: 0.1170 - val_loss: 0.1166\n",
      "Epoch 430/500\n",
      "34/34 - 7s - loss: 0.1155 - val_loss: 0.1140\n",
      "Epoch 431/500\n",
      "34/34 - 7s - loss: 0.1147 - val_loss: 0.1158\n",
      "Epoch 432/500\n",
      "34/34 - 7s - loss: 0.1149 - val_loss: 0.1160\n",
      "Epoch 433/500\n",
      "34/34 - 7s - loss: 0.1143 - val_loss: 0.1142\n",
      "Epoch 434/500\n",
      "34/34 - 7s - loss: 0.1139 - val_loss: 0.1145\n",
      "Epoch 435/500\n",
      "34/34 - 7s - loss: 0.1147 - val_loss: 0.1148\n",
      "Epoch 436/500\n",
      "34/34 - 7s - loss: 0.1137 - val_loss: 0.1134\n",
      "Epoch 437/500\n",
      "34/34 - 7s - loss: 0.1160 - val_loss: 0.1152\n",
      "Epoch 438/500\n",
      "34/34 - 7s - loss: 0.1135 - val_loss: 0.1159\n",
      "Epoch 439/500\n",
      "34/34 - 7s - loss: 0.1135 - val_loss: 0.1150\n",
      "Epoch 440/500\n",
      "34/34 - 7s - loss: 0.1144 - val_loss: 0.1151\n",
      "Epoch 441/500\n",
      "34/34 - 8s - loss: 0.1138 - val_loss: 0.1155\n",
      "Epoch 442/500\n",
      "34/34 - 7s - loss: 0.1142 - val_loss: 0.1176\n",
      "Epoch 443/500\n",
      "34/34 - 7s - loss: 0.1143 - val_loss: 0.1132\n",
      "Epoch 444/500\n",
      "34/34 - 7s - loss: 0.1130 - val_loss: 0.1141\n",
      "Epoch 445/500\n",
      "34/34 - 7s - loss: 0.1134 - val_loss: 0.1138\n",
      "Epoch 446/500\n",
      "34/34 - 7s - loss: 0.1145 - val_loss: 0.1132\n",
      "Epoch 447/500\n",
      "34/34 - 7s - loss: 0.1153 - val_loss: 0.1175\n",
      "Epoch 448/500\n",
      "34/34 - 7s - loss: 0.1128 - val_loss: 0.1137\n",
      "Epoch 449/500\n",
      "34/34 - 7s - loss: 0.1131 - val_loss: 0.1150\n",
      "Epoch 450/500\n",
      "34/34 - 7s - loss: 0.1130 - val_loss: 0.1144\n",
      "Epoch 451/500\n",
      "34/34 - 7s - loss: 0.1133 - val_loss: 0.1123\n",
      "Epoch 452/500\n",
      "34/34 - 7s - loss: 0.1136 - val_loss: 0.1136\n",
      "Epoch 453/500\n",
      "34/34 - 7s - loss: 0.1137 - val_loss: 0.1133\n",
      "Epoch 454/500\n",
      "34/34 - 7s - loss: 0.1136 - val_loss: 0.1147\n",
      "Epoch 455/500\n",
      "34/34 - 7s - loss: 0.1128 - val_loss: 0.1141\n",
      "Epoch 456/500\n",
      "34/34 - 7s - loss: 0.1135 - val_loss: 0.1134\n",
      "Epoch 457/500\n",
      "34/34 - 7s - loss: 0.1142 - val_loss: 0.1135\n",
      "Epoch 458/500\n",
      "34/34 - 7s - loss: 0.1128 - val_loss: 0.1136\n",
      "Epoch 459/500\n",
      "34/34 - 7s - loss: 0.1133 - val_loss: 0.1156\n",
      "Epoch 460/500\n",
      "34/34 - 7s - loss: 0.1140 - val_loss: 0.1145\n",
      "Epoch 461/500\n",
      "34/34 - 7s - loss: 0.1131 - val_loss: 0.1147\n",
      "Epoch 462/500\n",
      "34/34 - 7s - loss: 0.1131 - val_loss: 0.1144\n",
      "Epoch 463/500\n",
      "34/34 - 7s - loss: 0.1136 - val_loss: 0.1124\n",
      "Epoch 464/500\n",
      "34/34 - 8s - loss: 0.1138 - val_loss: 0.1117\n",
      "Epoch 465/500\n",
      "34/34 - 8s - loss: 0.1122 - val_loss: 0.1163\n",
      "Epoch 466/500\n",
      "34/34 - 7s - loss: 0.1130 - val_loss: 0.1133\n",
      "Epoch 467/500\n",
      "34/34 - 7s - loss: 0.1126 - val_loss: 0.1154\n",
      "Epoch 468/500\n",
      "34/34 - 8s - loss: 0.1129 - val_loss: 0.1131\n",
      "Epoch 469/500\n",
      "34/34 - 8s - loss: 0.1126 - val_loss: 0.1135\n",
      "Epoch 470/500\n",
      "34/34 - 7s - loss: 0.1142 - val_loss: 0.1149\n",
      "Epoch 471/500\n",
      "34/34 - 8s - loss: 0.1134 - val_loss: 0.1150\n",
      "Epoch 472/500\n",
      "34/34 - 8s - loss: 0.1117 - val_loss: 0.1133\n",
      "Epoch 473/500\n",
      "34/34 - 8s - loss: 0.1124 - val_loss: 0.1119\n",
      "Epoch 474/500\n",
      "34/34 - 8s - loss: 0.1115 - val_loss: 0.1159\n",
      "Epoch 475/500\n",
      "34/34 - 7s - loss: 0.1123 - val_loss: 0.1127\n",
      "Epoch 476/500\n",
      "34/34 - 7s - loss: 0.1127 - val_loss: 0.1122\n",
      "Epoch 477/500\n",
      "34/34 - 8s - loss: 0.1130 - val_loss: 0.1143\n",
      "Epoch 478/500\n",
      "34/34 - 7s - loss: 0.1113 - val_loss: 0.1133\n",
      "Epoch 479/500\n",
      "34/34 - 8s - loss: 0.1129 - val_loss: 0.1142\n",
      "Epoch 480/500\n",
      "34/34 - 8s - loss: 0.1129 - val_loss: 0.1142\n",
      "Epoch 481/500\n",
      "34/34 - 8s - loss: 0.1118 - val_loss: 0.1120\n",
      "Epoch 482/500\n",
      "34/34 - 8s - loss: 0.1120 - val_loss: 0.1111\n",
      "Epoch 483/500\n",
      "34/34 - 7s - loss: 0.1121 - val_loss: 0.1144\n",
      "Epoch 484/500\n",
      "34/34 - 7s - loss: 0.1125 - val_loss: 0.1153\n",
      "Epoch 485/500\n",
      "34/34 - 7s - loss: 0.1123 - val_loss: 0.1125\n",
      "Epoch 486/500\n",
      "34/34 - 8s - loss: 0.1131 - val_loss: 0.1142\n",
      "Epoch 487/500\n",
      "34/34 - 7s - loss: 0.1127 - val_loss: 0.1133\n",
      "Epoch 488/500\n",
      "34/34 - 8s - loss: 0.1124 - val_loss: 0.1138\n",
      "Epoch 489/500\n",
      "34/34 - 7s - loss: 0.1120 - val_loss: 0.1153\n",
      "Epoch 490/500\n",
      "34/34 - 7s - loss: 0.1125 - val_loss: 0.1132\n",
      "Epoch 491/500\n",
      "34/34 - 7s - loss: 0.1118 - val_loss: 0.1124\n",
      "Epoch 492/500\n",
      "34/34 - 7s - loss: 0.1123 - val_loss: 0.1112\n",
      "Epoch 493/500\n",
      "34/34 - 7s - loss: 0.1131 - val_loss: 0.1140\n",
      "Epoch 494/500\n",
      "34/34 - 7s - loss: 0.1123 - val_loss: 0.1135\n",
      "Epoch 495/500\n",
      "34/34 - 7s - loss: 0.1121 - val_loss: 0.1135\n",
      "Epoch 496/500\n",
      "34/34 - 7s - loss: 0.1134 - val_loss: 0.1119\n",
      "Epoch 497/500\n",
      "34/34 - 7s - loss: 0.1116 - val_loss: 0.1124\n",
      "Epoch 498/500\n",
      "34/34 - 7s - loss: 0.1128 - val_loss: 0.1134\n",
      "Epoch 499/500\n",
      "34/34 - 8s - loss: 0.1122 - val_loss: 0.1106\n",
      "Epoch 500/500\n",
      "34/34 - 7s - loss: 0.1121 - val_loss: 0.1149\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<tensorflow.python.keras.callbacks.History at 0x2b6ee50e6250>"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Compile the model with an ADAM optimiser and mean absolute error loss function\n",
    "model.compile(\n",
    "    optimizer=Adam(learning_rate=model.LR),\n",
    "    loss=\"mean_absolute_error\",\n",
    ")\n",
    "\n",
    "# Train the model for 500 epochs\n",
    "model.fit(\n",
    "    x=train_generator,\n",
    "    validation_data=test_generator,\n",
    "    callbacks=[tensorboard_callback, checkpoint_callback_best, checkpoint_callback_last],\n",
    "    verbose=2, epochs=500,\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}