- Source: games/connectx
- Kaggle: https://www.kaggle.com/jamesmcguigan/connectx-manually-coded-strategies
- Kaggle: https://www.kaggle.com/jamesmcguigan/connectx-alphabeta-minimax-with-numba
Writeup on a variety of approaches to competitive AI agent design, exploring classic game: Connect 4
- Random Agent
- Simple Rules Agent
- Minimax with AlphaBeta Pruning, Iterative Deepening and custom heuristic
- Object Oriented / Numpy implementation (slow)
- Bitboard Implementation (fast)
- Source: games/knights-isolation
- Report: games/knights-isolation/report.ipynb
Minimax adversarial search with alphaBeta pruning, iterative deepening, area heuristic, and persistent caching.
Monty Carlo Tree Search reinforcement learning
- Source: search/ant_colony
- Kaggle: https://www.kaggle.com/jamesmcguigan/ant-colony-optimization-algorithm
- Kaggle: https://www.kaggle.com/jamesmcguigan/kmeans-ant-colony-optimization
Ant Colony Solution the Travelling Salesman Problem
- Source: search/arc
- Kaggle: https://www.kaggle.com/jamesmcguigan/arc-geometry-solvers
- Kaggle: https://www.kaggle.com/jamesmcguigan/arc-solver-multimodel
- Dataset: https://github.com/fchollet/ARC
ARC can be seen as a general artificial intelligence benchmark, as a program synthesis benchmark, or as a psychometric intelligence test. It is targeted at both humans and artificially intelligent systems that aim at emulating a human-like form of general fluid intelligence.
Object model and frameworks for reasoning by analogy function solvers.
Brute force search of simple geometry and tessellation transformations with numpy.
XGBoost with a large multidimensional feature map was able to auto-solve a wide range of transformations.
AbstractSolver - Proof of Concept code using inspect.signature() to figure out all possible permutations of f(g(h(x))) implementing a IoC dependency injection solver.
- Source: search/pacman
A* Search Algorithms to navigate Pacman round a maze
- Source: puzzles/sudoku
- Kaggle: https://www.kaggle.com/jamesmcguigan/z3-sudoku-solver/
Sudoku Solver that can solve the World's Hardest Sudoku
- Source: puzzles/n_queens
- Kaggle: https://www.kaggle.com/jamesmcguigan/n-queens-92-solutions-in-prolog
This code solves the N-Queens problem in Prolog using the CLP(FD): Constraint Logic Programming over Finite Domain Library
This is a general purpose solver that can handle addition, subtraction, multiplication, integer division and raising to powers.