README.md

Label-Critic TM for Self Supervised Learning - Part of my PhD work

Implements the Self-Supervised Tsetlin Machine to learn classification tasks without ground truthes. The repository also include the classic Tsetlin Machine, Coalesced Tsetlin Machine, Convolutional Tsetlin Machine, Regression Tsetlin Machine, and Weighted Tsetlin Machine, with support for continuous features and drop clause. Self-Supervised-TM is written in Python with wrappers for C and CUDA-based clause evaluation and updating.

To challenge the proposed method, we set a data labels-guess game to run muliple rounds of self-supervised classification or clustering tasks on synthetic, semi-synthetic, and real datasets.

Architecture

To reproduce the guess game expirements on synthetic data, from examples, run:

python synthetic_lctm.py

To reproduce the guess game expirements on semi-synthetic data, from example, run:

python mnist_lctm_augmented.py

To reproduce the guess game expirements on MNIST data, from example, run:

python mnist_lctm.py

For cluster interpretability, from examples, run:

python mnist_interpret_clusters.py