Traveling Waves Encode the Recent Past and Enhance Sequence Learning

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Keller, T. Anderson, Muller, Lyle, Sejnowski, Terrence, Welling, Max
Natura: Preprint
Pubblicazione: 2023
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866909137459216384
author Keller, T. Anderson
Muller, Lyle
Sejnowski, Terrence
Welling, Max
author_facet Keller, T. Anderson
Muller, Lyle
Sejnowski, Terrence
Welling, Max
contents Traveling waves of neural activity have been observed throughout the brain at a diversity of regions and scales; however, their precise computational role is still debated. One physically inspired hypothesis suggests that the cortical sheet may act like a wave-propagating system capable of invertibly storing a short-term memory of sequential stimuli through induced waves traveling across the cortical surface, and indeed many experimental results from neuroscience correlate wave activity with memory tasks. To date, however, the computational implications of this idea have remained hypothetical due to the lack of a simple recurrent neural network architecture capable of exhibiting such waves. In this work, we introduce a model to fill this gap, which we denote the Wave-RNN (wRNN), and demonstrate how such an architecture indeed efficiently encodes the recent past through a suite of synthetic memory tasks where wRNNs learn faster and reach significantly lower error than wave-free counterparts. We further explore the implications of this memory storage system on more complex sequence modeling tasks such as sequential image classification and find that wave-based models not only again outperform comparable wave-free RNNs while using significantly fewer parameters, but additionally perform comparably to more complex gated architectures such as LSTMs and GRUs.
format Preprint
id arxiv_https___arxiv_org_abs_2309_08045
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Traveling Waves Encode the Recent Past and Enhance Sequence Learning
Keller, T. Anderson
Muller, Lyle
Sejnowski, Terrence
Welling, Max
Neural and Evolutionary Computing
Artificial Intelligence
Machine Learning
Traveling waves of neural activity have been observed throughout the brain at a diversity of regions and scales; however, their precise computational role is still debated. One physically inspired hypothesis suggests that the cortical sheet may act like a wave-propagating system capable of invertibly storing a short-term memory of sequential stimuli through induced waves traveling across the cortical surface, and indeed many experimental results from neuroscience correlate wave activity with memory tasks. To date, however, the computational implications of this idea have remained hypothetical due to the lack of a simple recurrent neural network architecture capable of exhibiting such waves. In this work, we introduce a model to fill this gap, which we denote the Wave-RNN (wRNN), and demonstrate how such an architecture indeed efficiently encodes the recent past through a suite of synthetic memory tasks where wRNNs learn faster and reach significantly lower error than wave-free counterparts. We further explore the implications of this memory storage system on more complex sequence modeling tasks such as sequential image classification and find that wave-based models not only again outperform comparable wave-free RNNs while using significantly fewer parameters, but additionally perform comparably to more complex gated architectures such as LSTMs and GRUs.
title Traveling Waves Encode the Recent Past and Enhance Sequence Learning
topic Neural and Evolutionary Computing
Artificial Intelligence
Machine Learning
url https://arxiv.org/abs/2309.08045