Oliver Eberle - Explainable structured machine learning in similarity, graph and transformer models
Presenter
January 12, 2023
Abstract
Recorded 12 January 2023. Oliver Eberle of Technische Universität Berlin presents "Explainable structured machine learning in similarity, graph and transformer models" at IPAM's Explainable AI for the Sciences: Towards Novel Insights Workshop.
Abstract: Many widely used models such as deep similarity models, GNNs, and Transformer models, are highly non-linear and structured in ways that challenge the extraction of meaningful explanations. This presentation will outline explanation techniques that consider the particular model structure in the framework of layer-wise relevance propagation. This motivates the step to go beyond standard explanations in terms of input features that result in second-order and higher-order attributions and to extend existing approaches for evaluating and visualizing explanation techniques to these new types of explanations. Using these methods, a selection of research use cases is presented, i.e. quantifying knowledge evolution in early modern times, studying gender bias in language models, and probing Transformer explanations during task-solving. This presentation thus highlights that a careful treatment of model structure in XAI can improve their faithfulness, result in better explanations and enable novel insights.
Learn more online at: http://www.ipam.ucla.edu/programs/workshops/explainable-ai-for-the-sciences-towards-novel-insights/