Measuring Per-Unit Interpretability at Scale Without Humans | Robust Machine Learning – Max Planck Institute for Intelligent Systems

Institute Homepage

Institute Homepage Sign In

Research Overview

Comparative Vision Science

Does CLIP's Generalization Performance Mainly Stem from High Train-Test Similarity?

In Search of Forgotten Domain Generalization

Partial Success in Closing the Gap between Human and Machine Vision

Foundations of Machine Learning

Cross-Entropy Is All You Need To Invert the Data Generating Process

Contrastive Learning Inverts the Data Generating Process

Interaction Asymmetry: A General Principle for Learning Composable Abstractions

Mechanistic Interpretability

Measuring Per-Unit Interpretability at Scale Without Humans

Exemplary Natural Images explain CNN Activations better than State-Of-The-Art Feature Visualizations

How Well do Feature Visualizations Support Causal Understanding of CNN Activations?

Robust Machine Learning

Measuring Per-Unit Interpretability at Scale Without Humans

Auto interpret

In this work, we introduce the first scalable method to measure per-unit interpretability in vision neural networks without human involvement. Through a large-scale psychophysics study, we demonstrate strong alignment with human judgments~[

].