Scalable Interconnect Learning in Boolean Networks

TL;DR

This paper introduces a scalable, trainable interconnect for Differentiable Boolean Logic Networks that maintains constant parameter count with increasing input width, along with pruning methods to reduce model size.

Contribution

It presents a novel, scalable interconnect design for Boolean networks and two effective pruning stages for model compression, enhancing scalability and efficiency.

Findings

Interconnect design scales to wider layers without increasing parameters.

Pruning methods significantly reduce model size while maintaining accuracy.

Proposed methods outperform baseline approaches in compression-accuracy trade-offs.

Abstract

Learned Differentiable Boolean Logic Networks (DBNs) already deliver efficient inference on resource-constrained hardware. We extend them with a trainable, differentiable interconnect whose parameter count remains constant as input width grows, allowing DBNs to scale to far wider layers than earlier learnable-interconnect designs while preserving their advantageous accuracy. To further reduce model size, we propose two complementary pruning stages: an SAT-based logic equivalence pass that removes redundant gates without affecting performance, and a similarity-based, data-driven pass that outperforms a magnitude-style greedy baseline and offers a superior compression-accuracy trade-off.