Bayesian Telephone: Memory Consolidation and Recall as Generative Processes

This is a code repo for Bayesian Telephone: Memory Consolidation and Recall as Generative Processes, a paper written for the Algorithms of the Mind course at Yale.

Key Files

shapes3d.ipynb contains the code used to implement a Hopfield network and variational autoencoder to model consolidation and recall of memory using the shapes3d dataset.

Bayesian hypothesis.ipynb contains the code used to implement a Bayesian system that infers latent parameters associated with images from a Mitsuba 3 constructed dataset.

true observation hypothesis.ipynb contains the code used to implement the true observation hypothesis mentioned in the paper.