Identifying Semantic Component for Robust Molecular Property Prediction

Zijian Li,Zunhong Xu,Ruichu Cai,Zhenhui Yang,Yuguang Yan,Zhifeng Hao,Guangyi Chen,Kun Zhang

2023-11-09

Abstract:Although graph neural networks have achieved great success in the task of molecular property prediction in recent years, their generalization ability under out-of-distribution (OOD) settings is still under-explored. Different from existing methods that learn discriminative representations for prediction, we propose a generative model with semantic-components identifiability, named SCI. We demonstrate that the latent variables in this generative model can be explicitly identified into semantic-relevant (SR) and semantic-irrelevant (SI) components, which contributes to better OOD generalization by involving minimal change properties of causal mechanisms. Specifically, we first formulate the data generation process from the atom level to the molecular level, where the latent space is split into SI substructures, SR substructures, and SR atom variables. Sequentially, to reduce misidentification, we restrict the minimal changes of the SR atom variables and add a semantic latent substructure regularization to mitigate the variance of the SR substructure under augmented domain changes. Under mild assumptions, we prove the block-wise identifiability of the SR substructure and the comment-wise identifiability of SR atom variables. Experimental studies achieve state-of-the-art performance and show general improvement on 21 datasets in 3 mainstream benchmarks. Moreover, the visualization results of the proposed SCI method provide insightful case studies and explanations for the prediction results. The code is available at: <a class="link-external link-https" href="https://github.com/DMIRLAB-Group/SCI" rel="external noopener nofollow">this https URL</a>.

Machine Learning,Artificial Intelligence,Quantitative Methods

What problem does this paper attempt to address?

The paper aims to address the issue of out-of-distribution (OOD) generalization in molecular property prediction, specifically focusing on improving the robustness of graph neural network (GNN)-based models. The authors propose a generative model called Semantic-Component Identifiable (SCI) model to identify semantic-relevant (SR) and semantic-irrelevant (SI) components within molecular structures. The key contributions and objectives of the paper can be summarized as follows:

1. **Problem Statement**: Despite the success of GNNs in molecular property prediction, their generalization ability under OOD settings remains underexplored. Existing methods often lack explicit guarantees for identifying SR components, leading to potential misidentification and poor generalization.

2. **Proposed Solution**:
- **SCI Model**: The authors propose the SCI model, which aims to explicitly identify SR and SI components. This is achieved through a generative model that incorporates a semantic-components identifiability mechanism.
- **Data Generation Process**: A hierarchical data generation process is formulated, splitting the latent space between atoms and molecules into SI substructures, SR substructures, and SR atom variables.
- **Identification Guarantees**: Theoretical proofs are provided to ensure the component-wise identifiability of SR atom variables and the block-wise identifiability of SR substructures under certain assumptions.
- **Regularization Techniques**: To redu