Explainable Structured Machine Learning in Similarity, Graph and Transformer Models

Explore explainable structured machine learning techniques in similarity, graph, and transformer models through this 53-minute conference talk. Delve into explanation methods that consider model structure within the layer-wise relevance propagation framework. Discover how to go beyond standard input feature explanations to achieve higher-order attributions and extend evaluation approaches for these new explanation types. Examine research use cases including quantifying knowledge evolution in early modern times, studying gender bias in language models, and probing Transformer explanations during task-solving. Gain insights into how careful treatment of model structure in explainable AI can improve faithfulness, enhance explanations, and enable novel discoveries in the field.

Oliver Eberle - Explainable structured machine learning in similarity, graph and transformer models