Abstract: Methods and apparatus are disclosed for interpreting a deep neural network (DNN) using a Semantic Coherence Analysis (SCA)-based interpretation technique. In embodiments, a multi-layered DNN that was trained for one task is analyzed using the SCA technique to select one layer in the DNN that produces salient features for another task. In embodiments, the DNN layers are tested with test samples labeled with a set of concept labels. The output features of a DNN layer are gathered and analyzed according to the concepts. In embodiments, the output is scored with a semantic coherence score, which indicates how well the layer separates the concepts, and one layer is selected from the DNN based on its semantic coherence score. In some embodiments, a support vector machine (SVM) or additional neural network may be added to the selected layer and trained to generate classification results based on the outputs of the selected layer.

Abstract: A system for selecting different aspects of data objects to be matched with similar aspects of other data objects. A user inputs a search data object and a value. A neural network computes features for the search object at multiple layers that correspond to different aspects of the object. A descriptor is generated for the search object from features output at a layer position of the neural network determined from the value. The descriptor is compared to corresponding descriptors for objects in a collection to select objects that include aspects similar to an aspect of the search object. The user can change the value to view different objects that include aspects similar to other aspects of the search object. Thus, the user can explore different aspects of an object to find objects that include aspects similar to the aspect of the object that the user is interested in.

Abstract: Methods and apparatus are disclosed for interpreting a deep neural network (DNN) using a Semantic Coherence Analysis (SCA)-based interpretation technique. In embodiments, a multi-layered DNN that was trained for one task is analyzed using the SCA technique to select one layer in the DNN that produces salient features for another task. In embodiments, the DNN layers are tested with test samples labeled with a set of concept labels. The output features of a DNN layer are gathered and analyzed according to the concepts. In embodiments, the output is scored with a semantic coherence score, which indicates how well the layer separates the concepts, and one layer is selected from the DNN based on its semantic coherence score. In some embodiments, a support vector machine (SVM) or additional neural network may be added to the selected layer and trained to generate classification results based on the outputs of the selected layer.