Cognitively Inspired Learning of Incremental Drifting Concepts

TL;DR

This paper introduces a biologically inspired deep learning model that incrementally learns new concepts in a continual setting, effectively mitigating catastrophic forgetting by using a memory system and pseudo-rehearsal based on neural theories.

Contribution

It combines Parallel Distributed Processing and Complementary Learning Systems theories to develop a model capable of incremental learning without interference, inspired by human cognition.

Findings

Model successfully learns new concepts without forgetting previous ones.

Uses pseudo-rehearsal to generate experience replay data.

Achieves continual learning with reduced catastrophic interference.

Abstract

Humans continually expand their learned knowledge to new domains and learn new concepts without any interference with past learned experiences. In contrast, machine learning models perform poorly in a continual learning setting, where input data distribution changes over time. Inspired by the nervous system learning mechanisms, we develop a computational model that enables a deep neural network to learn new concepts and expand its learned knowledge to new domains incrementally in a continual learning setting. We rely on the Parallel Distributed Processing theory to encode abstract concepts in an embedding space in terms of a multimodal distribution. This embedding space is modeled by internal data representations in a hidden network layer. We also leverage the Complementary Learning Systems theory to equip the model with a memory mechanism to overcome catastrophic forgetting through implementing pseudo-rehearsal. Our model can generate pseudo-data points for experience replay and accumulate new experiences to past learned experiences without causing cross-task interference.