SAL: Selective Adaptive Learning for Backpropagation-Free Training with Sparsification

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Liu, Fanping, Yang, Hua, Zou, Jiasi
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915760795811840
author Liu, Fanping
Yang, Hua
Zou, Jiasi
author_facet Liu, Fanping
Yang, Hua
Zou, Jiasi
contents Standard deep learning relies on Backpropagation (BP), which is constrained by biologically implausible weight symmetry and suffers from significant gradient interference within dense representations. To mitigate these bottlenecks, we propose Selective Adaptive Learning (SAL), a training method that combines selective parameter activation with adaptive area partitioning. Specifically, SAL decomposes the parameter space into mutually exclusive, sample-dependent regions. This decoupling mitigates gradient interference across divergent semantic patterns and addresses explicit weight symmetry requirements through our refined feedback alignment. Empirically, SAL demonstrates competitive convergence rates, leading to improved classification performance across 10 standard benchmarks. Additionally, SAL achieves numerical consistency and competitive accuracy even in deep regimes (up to 128 layers) and large-scale models (up to 1B parameters). Our approach is loosely inspired by biological learning mechanisms, offering a plausible alternative that contributes to the study of scalable neural network training.
format Preprint
id arxiv_https___arxiv_org_abs_2601_21561
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle SAL: Selective Adaptive Learning for Backpropagation-Free Training with Sparsification
Liu, Fanping
Yang, Hua
Zou, Jiasi
Machine Learning
Artificial Intelligence
Standard deep learning relies on Backpropagation (BP), which is constrained by biologically implausible weight symmetry and suffers from significant gradient interference within dense representations. To mitigate these bottlenecks, we propose Selective Adaptive Learning (SAL), a training method that combines selective parameter activation with adaptive area partitioning. Specifically, SAL decomposes the parameter space into mutually exclusive, sample-dependent regions. This decoupling mitigates gradient interference across divergent semantic patterns and addresses explicit weight symmetry requirements through our refined feedback alignment. Empirically, SAL demonstrates competitive convergence rates, leading to improved classification performance across 10 standard benchmarks. Additionally, SAL achieves numerical consistency and competitive accuracy even in deep regimes (up to 128 layers) and large-scale models (up to 1B parameters). Our approach is loosely inspired by biological learning mechanisms, offering a plausible alternative that contributes to the study of scalable neural network training.
title SAL: Selective Adaptive Learning for Backpropagation-Free Training with Sparsification
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2601.21561