A Network of Biologically Inspired Rectified Spectral Units (ReSUs) Learns Hierarchical Features Without Error Backpropagation

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Qin, Shanshan, Pughe-Sanford, Joshua L., Genkin, Alexander, Ozdil, Pembe Gizem, Greengard, Philip, Sengupta, Anirvan M., Chklovskii, Dmitri B.
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866908736918913024
author Qin, Shanshan
Pughe-Sanford, Joshua L.
Genkin, Alexander
Ozdil, Pembe Gizem
Greengard, Philip
Sengupta, Anirvan M.
Chklovskii, Dmitri B.
author_facet Qin, Shanshan
Pughe-Sanford, Joshua L.
Genkin, Alexander
Ozdil, Pembe Gizem
Greengard, Philip
Sengupta, Anirvan M.
Chklovskii, Dmitri B.
contents We introduce a biologically inspired, multilayer neural architecture composed of Rectified Spectral Units (ReSUs). Each ReSU projects a recent window of its input history onto a canonical direction obtained via canonical correlation analysis (CCA) of previously observed past-future input pairs, and then rectifies either its positive or negative component. By encoding canonical directions in synaptic weights and temporal filters, ReSUs implement a local, self-supervised algorithm for progressively constructing increasingly complex features. To evaluate both computational power and biological fidelity, we trained a two-layer ReSU network in a self-supervised regime on translating natural scenes. First-layer units, each driven by a single pixel, developed temporal filters resembling those of Drosophila post-photoreceptor neurons (L1/L2 and L3), including their empirically observed adaptation to signal-to-noise ratio (SNR). Second-layer units, which pooled spatially over the first layer, became direction-selective -- analogous to T4 motion-detecting cells -- with learned synaptic weight patterns approximating those derived from connectomic reconstructions. Together, these results suggest that ReSUs offer (i) a principled framework for modeling sensory circuits and (ii) a biologically grounded, backpropagation-free paradigm for constructing deep self-supervised neural networks.
format Preprint
id arxiv_https___arxiv_org_abs_2512_23146
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle A Network of Biologically Inspired Rectified Spectral Units (ReSUs) Learns Hierarchical Features Without Error Backpropagation
Qin, Shanshan
Pughe-Sanford, Joshua L.
Genkin, Alexander
Ozdil, Pembe Gizem
Greengard, Philip
Sengupta, Anirvan M.
Chklovskii, Dmitri B.
Neurons and Cognition
We introduce a biologically inspired, multilayer neural architecture composed of Rectified Spectral Units (ReSUs). Each ReSU projects a recent window of its input history onto a canonical direction obtained via canonical correlation analysis (CCA) of previously observed past-future input pairs, and then rectifies either its positive or negative component. By encoding canonical directions in synaptic weights and temporal filters, ReSUs implement a local, self-supervised algorithm for progressively constructing increasingly complex features. To evaluate both computational power and biological fidelity, we trained a two-layer ReSU network in a self-supervised regime on translating natural scenes. First-layer units, each driven by a single pixel, developed temporal filters resembling those of Drosophila post-photoreceptor neurons (L1/L2 and L3), including their empirically observed adaptation to signal-to-noise ratio (SNR). Second-layer units, which pooled spatially over the first layer, became direction-selective -- analogous to T4 motion-detecting cells -- with learned synaptic weight patterns approximating those derived from connectomic reconstructions. Together, these results suggest that ReSUs offer (i) a principled framework for modeling sensory circuits and (ii) a biologically grounded, backpropagation-free paradigm for constructing deep self-supervised neural networks.
title A Network of Biologically Inspired Rectified Spectral Units (ReSUs) Learns Hierarchical Features Without Error Backpropagation
topic Neurons and Cognition
url https://arxiv.org/abs/2512.23146