hahtorch 0.0.3

Hebbian/Anti-Hebbian Learning for Pytorch

Figure 1: HaH block for image classification DNNs.

Hebbian/Anti-Hebbian Learning for Pytorch

If you have questions you can contact metehancekic [at] ucsb [dot] edu

Pre-requisites

Install the dependencies

numpy==1.20.2 torch==1.10.2

How to install

We have a pypi module which can be installed simply with following command:

python3 -m pip install --index-url https://test.pypi.org/simple/ --no-deps hahtorch

Or one can clone the repository.

git clone [e-mail]:metehancekic/HaH.git

Experiments

We used CIFAR-10 image classification to show the effectiveness of our module. We train a VGG16 in standard fashion and train another VGG16 that contains HaHblocks with layer-wise HaHCost as a supplement.

CIFAR10 Image Classification with VGG16 model as Backbone

Figure 2: HaH VGG16, our proposed architecture for HaH training.