Neural Ising Machines via Unrolling and Zeroth-Order Training

Sam Reifenstein, Timothée Leleu · 2026

Read Neural Ising Machines via Unrolling and Zeroth-Order Training by Sam Reifenstein, Timothée Leleu (2026) 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-2602-00302.txt.

Description

Sam Reifenstein and Timothee Leleu propose neural Ising machines via unrolling and zeroth order training. Their data driven method tackles NP-hard Ising and Max-Cut optimization by learning an iterative dynamical system's update rule. Training uses a zeroth order optimizer, because backpropagation through long recurrent Ising dynamics gives unstable, poorly informative gradients.

Catalog record

CREATORSam Reifenstein, Timothée Leleu
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE62643
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2602.00302

Browse the full library · About the Tower of Babel