This repository contains code developed during a master's degree program in natural computing. The implemented algorithms focus on optimization and search techniques using artificial intelligence.
-
otimizacao.py: This notebook demonstrates an iterative approach of two AI algorithms for optimization and search: Hill Climbing and Simulated Annealing. It provides a step-by-step visualization of the algorithms using matplotlib. -
chromosome.py: An abstract class used for implementing various chromosome representations. Derived classes includechromosome_bitstring.py,chromosome_bitstring_max_function.py, andchromosome_simple_equation.py. -
genetic_algorithm.py: This file contains a general class for running genetic algorithms with support for any kind of chromosome representation. Therun()method has been modified multiple times to return generation age, fitness value, or the instance of the best chromosome. The structure of this code is inspired by David Kopec's approach outlined in the book "Classic Computer Science Problems in Python". -
main.py: Entry point for the codebase. -
Notebooks:
- Inside /plots
dispersao.ipynband3d_and_series.ipynbwere used for ad-hoc graph plotting and do not significantly contribute to the project's value. However, they are kept for future reference and potential reuse.
- Inside /plots
-
Neural Networks: This folder has the usage of the Perceptron architecture for classification of:
- file percpetron.ipynb - classification of the outputs of an logical port type AND
- folder perceptron constains a full Scala program to compute classifications on Iris, Wine and Aids dataset from scratch, also there are some Python notebooks for data analysis on the training results.
Here are some visualizations of the optimization process:
Visualization of the function ( f(x, y) = (1 − x)^2 + 100(y − x^2)^2 ), showcasing the results of the genetic algorithm optimization.
Evolution of the function result at each generation of the population initialized.
