Tensor methods have played a central role in analyzing multi-dimensional data across a wide range of real-world applications. At the same time, these methods provide low-rank representations, which enable trustworthy and parsimonious machine learning systems. However, even though they hold great promise, tensor methods remain relatively underexplored in the context of modern neural architectures and foundation models. This workshop aims to provide a forum for advancing tensor methods and applying them to various applications at the intersection of data mining and modern machine learning. Supported by organizers and keynote speakers with broad expertise across machine learning, signal processing, and data mining, the workshop aims to foster an interactive environment for researchers to exchange ideas and build connections across communities.
Dawon Ahn, Yong-chan Park, Taehyung Kwon et al.· Proceedings of the 32nd ACM...· 0 citations
We present ProgNet, a graph learning framework for interpretable graph classification that treats explanatory structures as first-class, reusable components of the prediction mechanism. Departing from existing methods that generate isolated, instance-specific explanations, ProgNet introduces a paradigm where reasoning is grounded in a shared vocabulary of reusable structural programs. Specifically, ProgNet represents each graph using a shared vocabulary of human-interpretable programs written in a graph pattern description language, grounding predictions in explicit structural evidence rather than latent embeddings alone. The vocabulary is constructed to promote both coverage and diversity, yielding compact and reusable structural primitives that generalize across instances. Classification is performed via an inherently decomposable evidence composition network that scores and aggregates program-level evidence, resulting in predictions whose logits admit additive, signed attributions. Extensive experiments on eight graph classification benchmarks demonstrate that ProgNet achieves competitive predictive accuracy while providing more faithful explanations.
Minseok Jeon, Seunghyun Park, Jun-Gi Jang· Proceedings of the 32nd ACM...· 0 citations