2020/09/12 by David Bau, Jun-Yan Zhu, Hendrik Strobelt +3 · 3 citations
Computer Science · #cs.CV #cs.LG #cs.NE #msc:68T07 #acm:68T07
paper · pdf · doi:10.1073/pnas.1907375117
Proceedings of the National Academy of Sciences 2020. Code at https://github.com/davidbau/dissect/ and website at https://dissect.csail.mit.edu/
arxiv created 2020/09/12 · arxiv updated 2020/09/15
Deep neural networks excel at finding hierarchical representations that solve complex tasks over large data sets. How can we humans understand these learned representations? In this work, we present network dissection, an analytic framework to systematically identify the semantics of individual hidden units within image classification and image generation networks. First, we analyze a convolutional neural network (CNN) trained on scene classification and discover units that match a diverse set of object concepts. We find evidence that the network has learned many object classes that play crucial roles in classifying scene classes. Second, we use a similar analytic method to analyze a generative adversarial network (GAN) model trained to generate scenes. By analyzing changes made when small sets of units are activated or deactivated, we find that objects can be added and removed from the output scenes while adapting to the context. Finally, we apply our analytic framework to understanding adversarial attacks and to semantic image editing.