Regression Concept Vectors for Bidirectional Explanations in Histopathology
Published in
(2018)External Links:
TL
TL;DR
This work proposes a methodology to exploit continuous concept measures as Regression Concept Vectors (RCVs) in the activation space of a layer to exploit network sensitivity to increasing values of a given concept measure.
Abstract
Explanations for deep neural network predictions in terms of domain-related concepts can be valuable in medical applications, where justifications are important for confidence in the decision-making. In this work, we propose a methodology to exploit continuous concept measures as Regression Concept Vectors (RCVs) in the activation space of a layer. The directional derivative of the decision function along the RCVs represents the network sensitivity to increasing values of a given concept measure. When applied to breast cancer grading, nuclei texture emerges as a relevant concept in the detection of tumor tissue in breast lymph node samples. We evaluate score robustness and consistency by statistical analysis.
References20 items
1
Cancer detection in histopathology whole-slide images using conditional random fields on deep embedded spaces
2
The (Un)reliability of saliency methods
3
Methods for interpreting and understanding deep neural networks
4
Understanding Black-box Predictions via Influence Functions
5