AlloyInter: Visualising Alloy Mixture Interpolations in t-SNE Representations

2025-09-23

AI Summary

Addressing the challenge of high-dimensional, nonlinear composition–property relationships in alloys—which hinder intuitive understanding and inverse design—this paper introduces AlloyInter, a bidirectional interactive exploration framework integrating model ensembling, eXplainable AI (XAI), and t-SNE manifold learning. Innovatively, XAI guides target-driven interpolation within the t-SNE embedding space, enabling semantically coherent, continuous transitions among alloy compositions while preserving local structure. The system supports user-specified target properties, enabling real-time inverse inference and visualization of optimal compositional pathways. Experimental evaluation demonstrates that AlloyInter achieves strong interpretability, robustness, and practical utility in multi-objective alloy design, significantly enhancing materials scientists’ comprehension and controllability over complex compositional spaces.

Technology Category

• Humans and AI: Explainable AI (XAI) for Human Understanding
• Machine Learning: Transparent, Interpretable, Explainable ML
• Search and Optimization: Mixed Discrete/Continuous Search

Application Category

• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

This entry description proposes AlloyInter, a novel system to enable joint exploration of input mixtures and output parameters space in the context of the SciVis Contest 2025. We propose an interpolation approach, guided by eXplainable Artificial Intelligence (XAI) based on a learned model ensemble that allows users to discover input mixture ratios by specifying output parameter goals that can be iteratively adjusted and improved towards a goal. We strengthen the capabilities of our system by building upon prior research within the robustness of XAI, as well as combining well-established techniques like manifold learning with interpolation approaches.

Problem

• Visualizing alloy mixture interpolations in t-SNE representations
• Discovering input mixture ratios by specifying output parameter goals
• Combining XAI-guided interpolation with manifold learning techniques

Innovation

• XAI-guided interpolation approach for alloy mixtures
• Model ensemble enables iterative parameter adjustment
• Combines manifold learning with interpolation techniques