Manuel Pfeuffer

ICON Decomposition: Auditing deep neural networks for shortcuts by decomposing layer-wise representations using concepts

Deep neural networks often exploit spurious associations, a failure known as shortcut learning. Before deployment, models should be audited for reliance on a set of concepts, such as acquisition artifacts or demographics. Current methods, such as linear probes and concept activation vectors, measure reliance by asking...

ICON Decomposition : Multivariate Concept-Level Explanations of Deep Representations for Model Auditing

Controlling for Omitted Variable Bias in Deep Neural Networks

This work introduces an estimation procedure that refits the final layer of a pre-trained network to include covariate effects, and shows how these effects can be orthogonalised with respect to covariates to exclude their mediated effects and that model predictions can be marginalised over the covariate distribution to...

ICON Decomposition: Auditing Deep Neural Networks with Multivariate Variance-based Concept-level Explanations

CON decomposition is introduced, which quantifies how much of a layer's variance each concept explains given all other concepts and the outcome, and how much none of them explains, yielding layer-comparable, calibrated scores that suppress false positives.

Interpreting Protein Language Model Embeddings via Orthogonal Projection for Protein Fitness Prediction

It is shown that PLM embeddings encode patterns correlated with biochemical properties and quantify their contribution to predicting protein fitness, and this technique is readily transferable to problem settings beyond protein fitness prediction.