Calibration and Generalizability of Probabilistic Models on Low-Data Chemical Datasets | Gary Tom
If you enjoyed this talk, consider joining the Molecular Modeling and Drug Discovery (M2D2) talks live: https://m2d2.io/talks/m2d2/about/
Also consider joining the M2D2 Slack: https://m2d2group.slack.com/join/shar...
Title: Calibration and generalizability of probabilistic models on low-data chemical datasets with DIONYSUS
Abstract: Deep learning models that leverage large datasets are often the state of the art for modelling molecular properties. When the datasets are smaller (less than 2000 molecules), it is not clear that deep learning approaches are the right modelling tool. In this work we perform an extensive study of the calibration and generalizability of probabilistic machine learning models on small chemical datasets. Using different molecular representations and models, we analyse the quality of their predictions and uncertainties in a variety of tasks (binary, regression) and datasets. We also introduce two simulated experiments that evaluate their performance: (1) Bayesian optimization guided molecular design, (2) inference on out-of-distribution data via ablated cluster splits. We offer practical insights into model and feature choice for modelling small chemical datasets, a common scenario in new chemical experiments. We have packaged our analysis into the DIONYSUS repository, which is open sourced to aid in reproducibility and extension to new datasets.
Speaker: Gary Tom - / mistergtom
Twitter Prudencio: / tossouprudencio
Twitter Therence: / therence_mtl
Twitter Jonny: / hsu_jonny
Twitter Valence Discovery: / valence_ai
~
Chapters:
00:00 - Intro
02:05 - Motivations
05:46 - Overview of Proposed Experiments
08:30 - Experiment 1: Study of Performance
29:44 - Experiment 2: Bayesian Optimization
41:18 - Experiment 3: Generalization and Ablation
47:46 - Practical Insights & Recommendations
50:30 - Q+A