A multi-objective Neural Search Algorithm for Capsule Neural Networks
Two set of environments are providided for mac m1 and linux users.
For pip users :
pip install -r <linux-requirement.txt> or <mac-requirements.txt>
For conda users :
conda env create -f <linux-environment.yml> or <linux-environment.yml>
We used a NSGA-II algorithm to optimise CapsNet on inference time and accuracy. Here are the results on MNIST dataset compared to our baselines models.
