Keywords: hopfield networks, modern hopfield networks, minimum description length, generalization, kolmogorov complexity, associative memory, memorization
TL;DR: We apply the Minimum Description Length principle to Hopfield networks to better optimize the tradeoff between generalization and memorization in these networks.
Abstract: Associative memory architectures are designed for memorization but also offer, through their retrieval method, a form of generalization to unseen inputs: stored memories can be seen as prototypes from this point of view. Focusing on Modern Hopfield Networks (MHN), we show that a large memorization capacity undermines the generalization opportunity. We offer a solution to better optimize this tradeoff. It relies on Minimum Description Length (MDL) to determine during training which memories to store, as well as how many of them.
Submission Number: 44
Loading