Predicting integers from continuous parameters
Bas Maat ∘, Peter Bloem ∘ · 2026
Read Predicting integers from continuous parameters by Bas Maat ∘, Peter Bloem ∘ (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-2602-10751.txt.
Description
Bas Maat and Peter Bloem study how to predict whole numbers from steady inputs in a 2026 paper. Labels like up-vote counts or ready rental bicycles are integers, yet standard practice models them as steady values. That switch quietly changes the label spread from discrete to steady, discarding structure.
The authors ask whether integer labels can be modeled directly by a discrete spread whose settings are predicted from instance traits. For neural network outputs, the settings must stay steady so backpropagation can learn the weights. The paper develops that direct discrete approach, keeping the count nature of the labels while keeping slope-based training intact. The work matters wherever predictions must be counts, not approximations of counts.
Catalog record
| CREATOR | Bas Maat ∘, Peter Bloem ∘ |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 101 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602-10751