What Makes a Good Feedforward Computational Graph?
Alex Vitvitskyi, João G.M. Araújo, Marc Lackenby, Petar Veličković · 2025
Read What Makes a Good Feedforward Computational Graph? by Alex Vitvitskyi, João G.M. Araújo, Marc Lackenby, Petar Veličković (2025) 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-2502-06751.txt.
Description
Alex Vitvitskyi, João Araújo, Marc Lackenby, and Petar Veličković wrote this paper in two thousand twenty-five. It asks what makes a feedforward computational graph, the wiring a neural network thinks through, a good one.
Graph choice can cripple a model through under-reaching or over-squashing, yet most study covers undirected graphs. For feedforward graphs, which have no back edges, the authors name two twin gauges: fidelity and mixing time. They prove how each grows for common graph shapes. They then train nets on those graphs and show the gauges track the scores. The pair gives designers a way to grade wiring before they train.
Catalog record
| CREATOR | Alex Vitvitskyi, João G.M. Araújo, Marc Lackenby, Petar Veličković |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 52 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2502-06751