LGSep 10, 2015

Use it or Lose it: Selective Memory and Forgetting in a Perpetual Learning Machine

arXiv:1509.03185v12 citations
Originality Incremental advance
AI Analysis

This work addresses the challenge of catastrophic forgetting in perpetual learning systems for AI applications, though it appears incremental as it builds on prior PLM concepts.

The authors tackled the problem of enabling deep neural networks to learn continuously like the brain by introducing selective memory and forgetting mechanisms during Perpetual Stochastic Gradient Descent, resulting in a 'use it or lose it' process where frequently recalled memories are retained and rarely recalled ones are forgotten, similar to human memory.

In a recent article we described a new type of deep neural network - a Perpetual Learning Machine (PLM) - which is capable of learning 'on the fly' like a brain by existing in a state of Perpetual Stochastic Gradient Descent (PSGD). Here, by simulating the process of practice, we demonstrate both selective memory and selective forgetting when we introduce statistical recall biases during PSGD. Frequently recalled memories are remembered, whilst memories recalled rarely are forgotten. This results in a 'use it or lose it' stimulus driven memory process that is similar to human memory.

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes