Learning Topology-Specific Experts for Molecular Property Prediction
نویسندگان
چکیده
Recently, graph neural networks (GNNs) have been successfully applied to predicting molecular properties, which is one of the most classical cheminformatics tasks with various applications. Despite their effectiveness, we empirically observe that training a single GNN model for diverse molecules distinct structural patterns limits its prediction performance. In this paper, motivated by observation, propose TopExpert leverage topology-specific models (referred as experts), each responsible group sharing similar topological semantics. That is, expert learns discriminative features while being trained corresponding group. To tackle key challenge grouping patterns, introduce clustering-based gating module assigns an input molecule into clusters and further optimizes two different types self-supervision: semantics induced GNNs scaffolds, respectively. Extensive experiments demonstrate has boosted performance property also achieved better generalization new unseen scaffolds than baselines. The code available at https://github.com/kimsu55/ToxExpert.
منابع مشابه
Learning the Experts for Online Sequence Prediction
Online sequence prediction is the problem of predicting the next element of a sequence given previous elements. This problem has been extensively studied in the context of individual sequence prediction, where no prior assumptions are made on the origin of the sequence. Individual sequence prediction algorithms work quite well for long sequences, where the algorithm has enough time to learn the...
متن کاملLearning and Prediction of Complex Molecular Structure- Property Relationships: Issues and Strategies for Modeling Intestinal Absorption for Drug Discovery
The problem of modeling and predicting complex structure-property relationships, such as the absorption, distribution, metabolism, and excretion of putative drug molecules is a fundamental one in contemporary drug discovery. An accurate model can not only be used to predict the behavior of a molecule and understand how structural variations may influence molecular property, but also to identify...
متن کاملBranch Meta-Prediction by Experts
Predicting the direction of conditional branches with high accuracy has become critical to the success of new deeper and wider processor pipelines. However, choosing a branch predictor is hard because the best choice is workload-dependent. Meta(or Hybrid-) predictors provide a rudimentary FSA-based heuristic for choosing between two predictors on-line. Such approaches have little or no theoreti...
متن کاملSoil Property Prediction: An Extreme Learning Machine Approach
In this paper, we propose a method for predicting functional properties of soil samples from a number of measurable spatial and spectral features of those samples. Our method is based on Savitzky-Golay filter for preprocessing and a relatively recent evolution of single hiddenlayer feed-forward network (SLFN) learning technique called extreme learning machine (ELM) for prediction. We tested our...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
ژورنال
عنوان ژورنال: Proceedings of the ... AAAI Conference on Artificial Intelligence
سال: 2023
ISSN: ['2159-5399', '2374-3468']
DOI: https://doi.org/10.1609/aaai.v37i7.26000