Synaptic Pruning: A Biological Inspiration for Deep Learning Regularization

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Vos, Gideon, van Eijk, Liza, Sarnyai, Zoltan, Azghadi, Mostafa Rahimi
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866916989718495232
author Vos, Gideon
van Eijk, Liza
Sarnyai, Zoltan
Azghadi, Mostafa Rahimi
author_facet Vos, Gideon
van Eijk, Liza
Sarnyai, Zoltan
Azghadi, Mostafa Rahimi
contents Synaptic pruning in biological brains removes weak connections to improve efficiency. In contrast, dropout regularization in artificial neural networks randomly deactivates neurons without considering activity-dependent pruning. We propose a magnitude-based synaptic pruning method that better reflects biology by progressively removing low-importance connections during training. Integrated directly into the training loop as a dropout replacement, our approach computes weight importance from absolute magnitudes across layers and applies a cubic schedule to gradually increase global sparsity. At fixed intervals, pruning masks permanently remove low-importance weights while maintaining gradient flow for active ones, eliminating the need for separate pruning and fine-tuning phases. Experiments on multiple time series forecasting models including RNN, LSTM, and Patch Time Series Transformer across four datasets show consistent gains. Our method ranked best overall, with statistically significant improvements confirmed by Friedman tests (p < 0.01). In financial forecasting, it reduced Mean Absolute Error by up to 20% over models with no or standard dropout, and up to 52% in select transformer models. This dynamic pruning mechanism advances regularization by coupling weight elimination with progressive sparsification, offering easy integration into diverse architectures. Its strong performance, especially in financial time series forecasting, highlights its potential as a practical alternative to conventional dropout techniques.
format Preprint
id arxiv_https___arxiv_org_abs_2508_09330
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Synaptic Pruning: A Biological Inspiration for Deep Learning Regularization
Vos, Gideon
van Eijk, Liza
Sarnyai, Zoltan
Azghadi, Mostafa Rahimi
Machine Learning
Artificial Intelligence
Synaptic pruning in biological brains removes weak connections to improve efficiency. In contrast, dropout regularization in artificial neural networks randomly deactivates neurons without considering activity-dependent pruning. We propose a magnitude-based synaptic pruning method that better reflects biology by progressively removing low-importance connections during training. Integrated directly into the training loop as a dropout replacement, our approach computes weight importance from absolute magnitudes across layers and applies a cubic schedule to gradually increase global sparsity. At fixed intervals, pruning masks permanently remove low-importance weights while maintaining gradient flow for active ones, eliminating the need for separate pruning and fine-tuning phases. Experiments on multiple time series forecasting models including RNN, LSTM, and Patch Time Series Transformer across four datasets show consistent gains. Our method ranked best overall, with statistically significant improvements confirmed by Friedman tests (p < 0.01). In financial forecasting, it reduced Mean Absolute Error by up to 20% over models with no or standard dropout, and up to 52% in select transformer models. This dynamic pruning mechanism advances regularization by coupling weight elimination with progressive sparsification, offering easy integration into diverse architectures. Its strong performance, especially in financial time series forecasting, highlights its potential as a practical alternative to conventional dropout techniques.
title Synaptic Pruning: A Biological Inspiration for Deep Learning Regularization
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2508.09330