Understanding Regularization to Visualize Convolutional Neural Networks
This work addresses the challenge of interpreting neural networks for researchers and practitioners, but it is incremental as it builds on existing regularization approaches.
The authors tackled the problem of visualizing concepts learned by convolutional neural networks by proposing a mathematical framework that unifies existing regularization methods and introduces a novel technique based on Sobolev gradients, resulting in sharper reconstructions and better control over scales in experiments.
Variational methods for revealing visual concepts learned by convolutional neural networks have gained significant attention during the last years. Being based on noisy gradients obtained via back-propagation such methods require the application of regularization strategies. We present a mathematical framework unifying previously employed regularization methods. Within this framework, we propose a novel technique based on Sobolev gradients which can be implemented via convolutions and does not require specialized numerical treatment, such as total variation regularization. The experiments performed on feature inversion and activation maximization demonstrate the benefit of a unified approach to regularization, such as sharper reconstructions via the proposed Sobolev filters and a better control over reconstructed scales.