Beyond Weighted Summation: Learnable Nonlinear Aggregation Functions for Robust Artificial Neurons
Berke Deniz Bozyigit · 2026
Read Beyond Weighted Summation: Learnable Nonlinear Aggregation Functions for Robust Artificial Neurons by Berke Deniz Bozyigit (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-2603-19344.txt.
Description
Bozyigit questions the fixed weighted sum at the heart of the classic artificial neuron. That sum acts like a mean, so noisy or extreme inputs can sway it. Two learnable swaps are offered: a power-weighted F-Mean rule and a distance-based Gaussian rule. Hybrid neurons blend the new rules with the linear sum through a learned dial. On CIFAR-10 and a noisy twin set, hybrids resist noise far better than the baseline. Scores reach 0.991 against 0.890 for standard neurons. Learned dials settle on sub-linear sums in most runs.
Catalog record
| CREATOR | Berke Deniz Bozyigit |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 20374 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2603.19344