Spike-Timing-Dependent Plasticity for Bernoulli Message Passing

Sepideh Adamiat, Wouter M. Kouw, Bert de Vries · 2025

Read Spike-Timing-Dependent Plasticity for Bernoulli Message Passing by Sepideh Adamiat, Wouter M. Kouw, Bert de Vries (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2512-23728.txt.

Description

Sepideh Adamiat, Wouter Kouw, and Bert de Vries design spiking neural networks that perform Bayesian inference the way brains might. Bayesian reasoning gives a principled account of perception and learning, while real neurons communicate in spikes, not in the smooth numbers of textbook inference. The paper connects the two pictures through message passing with Bernoulli messages, the simplest kind, each representing a binary probability.

Learning in their networks uses spike-timing-dependent plasticity, a biologically plausible rule in which synapses strengthen or weaken according to the relative timing of spikes, echoing Hebb's principle. The trained networks' behavior closely matched the true numerical solutions of the inference problems. As a versatility check, the authors implemented a factor graph from coding theory, showing the same spiking machinery can transmit signals over an unreliable channel.

Catalog record

CREATORSepideh Adamiat, Wouter M. Kouw, Bert de Vries
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE26 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2512-23728

Browse the full library · About the Tower of Babel