All projects16 / 16
Feature Selection with PSO & GA
Metaheuristics for dimensionality reduction.
A comparative implementation of Particle Swarm Optimisation and Genetic Algorithms applied to feature selection on high-dimensional datasets.
The problem
Exhaustive feature search is combinatorially impossible; metaheuristics trade guaranteed optimality for tractable, near-optimal subsets.
The approach
- 01
Implemented PSO and GA search over the feature space with a shared fitness interface.
- 02
Benchmarked selected subsets against baseline classifier performance.
Stack
- Core
- Python · NumPy · scikit-learn · Jupyter