Linearized Bregman Iterations for Sparse Spiking Neural Networks
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2026
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866911522479931392 |
|---|---|
| author | Windhager, Daniel Moser, Bernhard A. Lunglmayr, Michael |
| author_facet | Windhager, Daniel Moser, Bernhard A. Lunglmayr, Michael |
| contents | Spiking Neural Networks (SNNs) offer an energy efficient alternative to conventional Artificial Neural Networks (ANNs) but typically still require a large number of parameters. This work introduces Linearized Bregman Iterations (LBI) as an optimizer for training SNNs, enforcing sparsity through iterative minimization of the Bregman distance and proximal soft thresholding updates. To improve convergence and generalization, we employ the AdaBreg optimizer, a momentum and bias corrected Bregman variant of Adam. Experiments on three established neuromorphic benchmarks, i.e. the Spiking Heidelberg Digits (SHD), the Spiking Speech Commands (SSC), and the Permuted Sequential MNIST (PSMNIST) datasets, show that LBI based optimization reduces the number of active parameters by about 50% while maintaining accuracy comparable to models trained with the Adam optimizer, demonstrating the potential of convex sparsity inducing methods for efficient neuromorphic learning. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_16462 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Linearized Bregman Iterations for Sparse Spiking Neural Networks Windhager, Daniel Moser, Bernhard A. Lunglmayr, Michael Signal Processing Neural and Evolutionary Computing Spiking Neural Networks (SNNs) offer an energy efficient alternative to conventional Artificial Neural Networks (ANNs) but typically still require a large number of parameters. This work introduces Linearized Bregman Iterations (LBI) as an optimizer for training SNNs, enforcing sparsity through iterative minimization of the Bregman distance and proximal soft thresholding updates. To improve convergence and generalization, we employ the AdaBreg optimizer, a momentum and bias corrected Bregman variant of Adam. Experiments on three established neuromorphic benchmarks, i.e. the Spiking Heidelberg Digits (SHD), the Spiking Speech Commands (SSC), and the Permuted Sequential MNIST (PSMNIST) datasets, show that LBI based optimization reduces the number of active parameters by about 50% while maintaining accuracy comparable to models trained with the Adam optimizer, demonstrating the potential of convex sparsity inducing methods for efficient neuromorphic learning. |
| title | Linearized Bregman Iterations for Sparse Spiking Neural Networks |
| topic | Signal Processing Neural and Evolutionary Computing |
| url | https://arxiv.org/abs/2603.16462 |