An Adaptive Strategy Particle Swarm Optimiser for CNN Training
The code for the results presented in "Large Scale Optimisation with Adaptive Strategy Particle Swarms" by Ali Shihab. The notebook has been set-up to be run sequentially and will automatically choose between a cuda c++ enabled GPU or an Apple Silicon chip, depending on what is available, by preference for the cuda device. CPU is the fall back device, but this may cause unexpected behaviour and is not recommended.
-
pip install -r requirements.txt -
ensure that you have the necessary directories set up in order to "saving" portion of the code, otherwise you can adapt the code to save at your specified location, or you can choose not to run the saving portion of the code at all.
-
enjoy
Placed among the top of my cohort despite being the only single-member team (all others were 3-member teams)