Switchable Activation Networks
Laha Ale, Ning Zhang, Scott A. King, Pingzhi Fan · 2026
Read Switchable Activation Networks by Laha Ale, Ning Zhang, Scott A. King, Pingzhi Fan (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-06601.txt.
Description
Ale and team bring SWAN, a tool giving each neural unit with an input-dependent binary gate, learning when each unit should be active or inactive for work allocation. Deep networks and large generative models achieve performance yet carry cost blocking resource-tight deployment. Dropout steadies training, while pruning and low-rank splitting turn models post hoc into static forms with flex. Unlike pruning, SWAN learns context-set activation patterns backing efficient dynamic inference and switch to compact dense models, joining sparsity, pruning, and adaptive use in one plan.
Catalog record
| CREATOR | Laha Ale, Ning Zhang, Scott A. King, Pingzhi Fan |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 41885 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2603.06601