Summary of Retrosynthesis Prediction Enhanced by In-silico Reaction Data Augmentation, By Xu Zhang and Yiming Mo and Wenguan Wang and Yi Yang
Retrosynthesis prediction enhanced by in-silico reaction data augmentation
by Xu Zhang, Yiming Mo, Wenguan Wang, Yi Yang
First submitted to arxiv on: 31 Jan 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary This research paper presents a novel framework called RetroWISE that addresses the limitations of machine learning (ML) methods for retrosynthesis research. Currently, these methods require substantial amounts of paired training data, which is costly to obtain and often restricted by companies. The authors propose exploiting unpaired data to generate in-silico paired data, enabling more efficient model training. Specifically, RetroWISE uses a base model trained on real paired data to generate and augment reactions using unpaired data, ultimately leading to superior models. Experimental results demonstrate that RetroWISE achieves the best overall performance on three benchmark datasets, including the USPTO-50K test dataset (+8.6% top-1 accuracy), and consistently improves the prediction accuracy of rare transformations. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This research is all about making it easier for scientists to design new chemical reactions using machines. Right now, computers need a lot of data to learn how to do this efficiently, but that data can be hard to get. The authors came up with an idea: what if we use smaller pieces of information to generate the missing parts? They created a special system called RetroWISE that does just that. It takes some initial training data and uses it to make new reactions, which helps it learn even better. The results are impressive – their method is the best on several tests and can predict rare reactions more accurately. |
Keywords
* Artificial intelligence * Machine learning
