Helmholtz Imaging CONNECT

Go back Home Solutions Multi-path Explanations Library

Multi-path Explanations Library (m-pax_lib)

Authors: Lukas Klein, João B. S. Carvalho

Keywords: Explanatory AI, Disentangled Representations, Shortcut Detection, Medical Imaging

This repository contains a framework which allows for enhanced model interpretability and model robustness. The framework utilizes interpretable disentangled representations for downstream-task prediction which are in a second step visualized and evaluated through a multipath-attribution mapping.

Visualizing and interpreting disentangled representations through multipath-attribution maps.

©Lukas Klein

License

Apache License 2.0

Multi-path Explanations Library in Helmholtz Imaging CONNECT:

Expert

Lukas Klein

Deutsches Krebsforschungszentrum - Heidelberg

Skills:
Explanatory Methods, Interpretable Representation-learning, Model uncertainty, Probabilistic AI, Bayesian NN
Projects:
Improving Explainability through Disentangled Feature Representation, XAI in Multiple Instance Learning, Explaining Model-Uncertainty

In Helmholtz Imaging CONNECT
➤Modalities: 0 ➤Applications: 2 ➤Instruments: 0 ➤Labs: 2 ➤Facilities: 0 ➤Centers: 1 ➤Solutions: 1

Expert

Helmholtz Imaging Support

Helmholtz Imaging offers support for any imaging challenge, independent of the modality. If you do not find the expert here you are looking for, please don't hesitate to contact us!

More than 30 imaging experts are ready to help you.

Application

Explainable AI

Explainable AI (XAI) aims to render model behavior understandable by humans, which can be seen as an intermediate step in extracting causal relations from correlative patterns. Due to the high risk of possible fatal decisions …

In Helmholtz Imaging CONNECT
➤Facilities: 0 ➤Centers: 5 ➤Instruments: 0 ➤Labs: 0 ➤Modalities: 0 ➤Experts: 5 ➤Solutions: 1

No facility found.

Center

Deutsches Krebsforschungszentrum (DKFZ)

More than 450,000 people are diagnosed with cancer each year in Germany. Cancer is a disease that poses enormous challenges to research, because every cancer is different and its course can vary immensely even from …

In Helmholtz Imaging CONNECT
➤Applications: 40 ➤Facilities: 0 ➤Instruments: 6 ➤Labs: 17 ➤Modalities: 31 ➤Experts: 29 ➤Solutions: 9

No instrument found.

No modality found.

Lab

Interactive Machine Learning (IML)

Considering human interaction when designing machine learning (ML) systems bears great potential: On the one hand, decision-making in ML systems remains imperfect in practice, thus requiring human interaction for safety-critical applications such as clinical diagnostics. …

In Helmholtz Imaging CONNECT
➤Applications: 0 ➤Facilities: 0 ➤Centers: 2 ➤Instruments: 0 ➤Modalities: 0 ➤Experts: 4 ➤Solutions: 2