Internal Examiner
: · 2026
Read Internal Examiner by : (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-2602-04078.txt.
Description
Róisín Luo's 2026 doctoral thesis asks why neural networks resist small attacks and generalize to new data. The central tool is Lipschitz continuity, a bound on how much a network's output can change when its input changes slightly. That worst-case sensitivity governs both robustness and generalization in theory.
Earlier work mostly used Lipschitz ideas through empirical regularization tricks. Luo instead pursues the underlying principles directly, asking what the bound means for how networks should be designed and trained. The thesis aims to turn a widely used constraint into understood science, so robustness can be engineered rather than stumbled upon.
Catalog record
| CREATOR | : |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 42 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602-04078