Visualizing Deep Neural Network Decisions: Prediction Difference Analysis

Authors: Luisa M Zintgraf, Taco S Cohen, Tameem Adel, Max Welling

ICLR 2017 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We illustrate the method in experiments on natural images (Image Net data), as well as medical images (MRI brain scans).
Researcher Affiliation Academia 1University of Amsterdam, 2Canadian Institute of Advanced Research, 3Vrije Universiteit Brussel {lmzintgraf,tameem.hesham}@gmail.com, {t.s.cohen, m.welling}@uva.nl
Pseudocode Yes Algorithm 1 Evaluating the prediction difference using conditional and multivariate sampling
Open Source Code Yes Our implementation is available at github.com/lmzintgraf/Deep Vis-Pred Diff.
Open Datasets Yes We use images from the ILSVRC challenge (Russakovsky et al., 2015) (a large dataset of natural images from 1000 categories)
Dataset Splits Yes to achieve an accuracy of 69.3% in a 10-fold cross-validation test.
Hardware Specification No No specific hardware details like exact GPU/CPU models or memory amounts used for computation were provided. The paper mentions 'using the GPU implementation of caffe' and 'on a CPU', and describes MRI scanner hardware: 'Subjects were scanned on two 3.0 Tesla scanner systems, 121 subjects on a Philips Intera system and 39 on a Philips Ingenia system.'
Software Dependencies No The paper mentions using 'caffe' and 'software developed in-house' for preprocessing, but does not specify version numbers for these or any other key libraries/solvers.
Experiment Setup Yes mini-batches with the standard settings of 10 samples and a window size of k = 10.