Saved in:
Bibliographic Details
Main Authors: Ale, Laha, Zhang, Ning, King, Scott A., Fan, Pingzhi
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2603.06601
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915840100663296
author Ale, Laha
Zhang, Ning
King, Scott A.
Fan, Pingzhi
author_facet Ale, Laha
Zhang, Ning
King, Scott A.
Fan, Pingzhi
contents Deep neural networks, and more recently large-scale generative models such as large language models (LLMs) and large vision-action models (LVAs), achieve remarkable performance across diverse domains, yet their prohibitive computational cost hinders deployment in resource-constrained environments. Existing efficiency techniques offer only partial remedies: dropout improves regularization during training but leaves inference unchanged, while pruning and low-rank factorization compress models post hoc into static forms with limited adaptability. Here we introduce SWAN (Switchable Activation Networks), a framework that equips each neural unit with a deterministic, input-dependent binary gate, enabling the network to learn when a unit should be active or inactive. This dynamic control mechanism allocates computation adaptively, reducing redundancy while preserving accuracy. Unlike traditional pruning, SWAN does not simply shrink networks after training; instead, it learns structured, context-dependent activation patterns that support both efficient dynamic inference and conversion into compact dense models for deployment. By reframing efficiency as a problem of learned activation control, SWAN unifies the strengths of sparsity, pruning, and adaptive inference within a single paradigm. Beyond computational gains, this perspective suggests a more general principle of neural computation, where activation is not fixed but context-dependent, pointing toward sustainable AI, edge intelligence, and future architectures inspired by the adaptability of biological brains.
format Preprint
id arxiv_https___arxiv_org_abs_2603_06601
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Switchable Activation Networks
Ale, Laha
Zhang, Ning
King, Scott A.
Fan, Pingzhi
Machine Learning
Deep neural networks, and more recently large-scale generative models such as large language models (LLMs) and large vision-action models (LVAs), achieve remarkable performance across diverse domains, yet their prohibitive computational cost hinders deployment in resource-constrained environments. Existing efficiency techniques offer only partial remedies: dropout improves regularization during training but leaves inference unchanged, while pruning and low-rank factorization compress models post hoc into static forms with limited adaptability. Here we introduce SWAN (Switchable Activation Networks), a framework that equips each neural unit with a deterministic, input-dependent binary gate, enabling the network to learn when a unit should be active or inactive. This dynamic control mechanism allocates computation adaptively, reducing redundancy while preserving accuracy. Unlike traditional pruning, SWAN does not simply shrink networks after training; instead, it learns structured, context-dependent activation patterns that support both efficient dynamic inference and conversion into compact dense models for deployment. By reframing efficiency as a problem of learned activation control, SWAN unifies the strengths of sparsity, pruning, and adaptive inference within a single paradigm. Beyond computational gains, this perspective suggests a more general principle of neural computation, where activation is not fixed but context-dependent, pointing toward sustainable AI, edge intelligence, and future architectures inspired by the adaptability of biological brains.
title Switchable Activation Networks
topic Machine Learning
url https://arxiv.org/abs/2603.06601