A Reconfigurable Stream-Based FPGA Accelerator for Bayesian Confidence Propagation Neural Networks
Muhammad Ihsan Al Hafiz, Naresh Ravichandran, Anders Lansner, Pawel Herman, Artur Podobas · 2025
Read A Reconfigurable Stream-Based FPGA Accelerator for Bayesian Confidence Propagation Neural Networks by Muhammad Ihsan Al Hafiz, Naresh Ravichandran, Anders Lansner, Pawel Herman, Artur Podobas (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-2503-01561.txt.
Description
Muhammad Ihsan Al Hafiz and four co-authors put a brain-style learning network onto reconfigurable hardware that streams data through it. Bayesian confidence propagation networks learn with local rules, without the global error signals of deep learning, and they can learn unsupervised with flexible structure. Those traits promise speed and low energy, but only if hardware runs them natively. Their 2025 paper designs an FPGA accelerator, a chip whose circuits are rewired for the task, fed by a stream of data rather than batches.
Brain-inspired algorithms often die as slow software demos; an accelerator is their proof of life. Streaming suits always-on sensors that cannot wait to batch. The authors target machine learning and neuroscience users of this network family. Their design is reconfigurable, so the same board serves varied models. The work is computer engineering in service of a learning theory.
Catalog record
| CREATOR | Muhammad Ihsan Al Hafiz, Naresh Ravichandran, Anders Lansner, Pawel Herman, Artur Podobas |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 39 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2503-01561