Characterizing Inter-Layer Functional Mappings of Deep Learning Models
This work addresses the need for interpretability in deep learning models, providing tools to analyze layer-specific contributions, though it is incremental as it applies existing statistical methods to new contexts.
The paper tackled the problem of understanding how deep learning models transform data across layers by using the Henze-Penrose statistic to characterize class separation distributions, demonstrating its utility on datasets like CIFAR10 and MNIST to analyze layer adaptation and contributions.
Deep learning architectures have demonstrated state-of-the-art performance for object classification and have become ubiquitous in commercial products. These methods are often applied without understanding (a) the difficulty of a classification task given the input data, and (b) how a specific deep learning architecture transforms that data. To answer (a) and (b), we illustrate the utility of a multivariate nonparametric estimator of class separation, the Henze-Penrose (HP) statistic, in the original as well as layer-induced representations. Given an $N$-class problem, our contribution defines the $C(N,2)$ combinations of HP statistics as a sample from a distribution of class-pair separations. This allows us to characterize the distributional change to class separation induced at each layer of the model. Fisher permutation tests are used to detect statistically significant changes within a model. By comparing the HP statistic distributions between layers, one can statistically characterize: layer adaptation during training, the contribution of each layer to the classification task, and the presence or absence of consistency between training and validation data. This is demonstrated for a simple deep neural network using CIFAR10 with random-labels, CIFAR10, and MNIST datasets.