Interpretable Deep Learning for Polar Mechanistic Reaction Prediction
Ryan J. Miller, Alexander E. Dashuta, Brayden Rudisill, David Van Vranken, Pierre Baldi · 2025
Read Interpretable Deep Learning for Polar Mechanistic Reaction Prediction by Ryan J. Miller, Alexander E. Dashuta, Brayden Rudisill, David Van Vranken, Pierre Baldi (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-2504-15539.txt.
Description
Ryan Miller and co-authors present a 2025 paper on predicting chemical reactions with deep learning. Reaction guesses drive new work in medicine, factories, and farming. Human solving is slow and costly. Most past models train on patent data and map reactants straight to products. That hides the step-by-step electron flow that chemists care about. The team trains on a data set built from polar elementary steps instead.
Their system, called PMechRP, learns mechanism steps with electron flow detail. Each step is small and readable, so the full path can be checked. Extra data work widens the range of reactions covered. Synthetic chemists get predictions they can inspect, not just accept. A readable mechanism lets a chemist spot a bad step before wasting lab time. The paper argues that mechanism-level training is the path to models that bench chemists will trust.
Catalog record
| CREATOR | Ryan J. Miller, Alexander E. Dashuta, Brayden Rudisill, David Van Vranken, Pierre Baldi |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 40 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2504-15539