A collection of different labs completed for my Artificial Intelligence class (6.034). Specifically, I cover representations, methods, and architectures used to build applications, and to account for understanding and developing intelligence from a computational perspective - aka "artificial intelligence" (AI).
Exploration of different graph search approaches and algorithms.
-
Blind Search
◦ Depth-First Search
◦ Breadth-First Search -
Heuristic Search
◦ Hill Climbing
◦ Best-First Search
◦ Beam Search -
Optimal Search
◦ Branch & Bound
◦ Branch & Bound w/ A*
Representing a game as a decision tree and performing search operations on the corresponding tree.
-
Representation Methods
◦ Game Completion
◦ Move Generation
◦ Determining Heuristics -
Search Methods
◦ Cooperative Depth-First Search
◦ Ordinary Minimax Search
◦ Minimax Search w/ Alpha-beta Pruning
◦ Progressive Deepening
Exploring different approaches to solve constraint satisfaction problems (CSPs).
-
Forward Checking
◦ Neighbor Elimination
◦ DFS Solver -
Propagation
◦ Domain Reduction
◦ Generic Propagation
◦ Propagation through Singleton domains
◦ Propagation through Reduced domains
Exploring the structure and application of production rule systems w/ data and rules.
- Goal Trees
◦ Family Relations
◦ Backward Chaining
Understanding the different components of building and evaluating Bayesian networks.
-
Probabilities
◦ Joint Prob.
◦ Marginal Prob.
◦ Conditional Prob. -
Independence
◦ Conditional Indep.
◦ Structurally Indep. -
Misc.
◦ # of Parameters
◦ Node Relationships
Understanding how to effectively classify and interpret information w/ ID trees and kNNs.
-
ID Trees
◦ Point Classification
◦ Disorder Calculation
◦ Tree Construction
◦ Greedy Construction -
k-Nearest Neighbors
◦ Distance Metrics
◦ Neighbor Classification
◦ k-Selection and Validation
Understanding how network models interpret input information and produce meaningful output.
-
Forward Propagation
◦ Node Value Determination
◦ Binary Output Calculation -
Backward Propagation
◦ Gradient Descent
◦ Determine Dependencies
◦ Delta-B Calculation
◦ Weight Update
Exploring another method of supervised point classification, with SVMs utilizing relevant training points.
-
Boundaries
◦ Positiveness
◦ Classification
◦ Margin
◦ Gutter Constraint -
Supportiveness
◦ Validation of Alpha Sign and Equation -
Evaluation
◦ Accuracy
◦ Training/SVM Update
Implementing the adaboost algorithm to combine the strengths of different classifiers.
- Adaboost Methods
◦ Weight Initialization
◦ Weight Update
◦ Error Calculation
◦ Voting Power
◦ Determine Misclassifications
◦ Algorithm for N Rounds