xAI application areas such as computer vision and natural language processing benefit from pre-existing human intuitions that guide the validation and understanding of model behaviors. In this special track, we propose a shift in focus for xAI toward domains where human intuition is limited or non-existent, particularly at the cutting of scientific research such as in chemistry and materials science. By leveraging xAI to interpret high-performing models in these areas, we open pathways to novel scientific insights and a deeper understanding of target properties. This special track seeks contributions that explore the unique demands and challenges of using xAI to generate scientific understanding in chemistry, materials science, and related fields. Submissions should address critical issues such as identifying effective explanation modalities to derive property-related insights, ensuring explanations accurately reflect underlying properties of the data distribution rather than model artifacts and biases, and developing methods to extract knowledge that is experimentally or theoretically verifiable.
- xAI-guided discovery of structure-property relationships in molecular and material datasets.
- Counterfactual explanations to predict minimal changes for property optimization in molecules.
- Techniques for ensuring the fidelity of explanations in predicting functional material behaviors.
- Model-agnostic frameworks for explaining molecular properties with structural descriptors.
- Adaptive xAI strategies for addressing biases in material property prediction models.
- Real-world case studies showcasing xAI applications in sustainable material innovations.
- Educational xAI applications for enhancing understanding of complex material science concepts.
- Integrating large language models with xAI for hypothesis generation in chemistry and materials science.
- Experimentally verifiable predictions derived from xAI-guided material design strategies.
- Employing xAI for autonomous systems in closed-loop simulations and experiments.
- Role of xAI in automating experimental setup suggestions for material discovery.
- Explanation methods for identifying functional substructures in drug discovery applications.
- Using xAI to uncover the role of molecular aggregation and other relevant properties in pharmaceutical applications.
- Visual explanation tools for exploring quantum effects in material properties.
- Domain-specific evaluation frameworks for validating xAI insights in quantum chemistry.
- Enhancing chemical reaction modeling through attributional explanations in neural networks.
- Combining xAI and machine learning to predict reaction mechanisms in catalysis studies.
- Multi-scale explanation techniques for bridging atomic and bulk property predictions.
- Hybrid visual-textual explanation methods for analyzing crystal structures and defects.
- xAI tools for visualizing structure-performance links in energy storage materials.
