Concept Whitening for Interpretable Image Recognition
This addresses interpretability challenges in machine learning for researchers and practitioners, offering a novel method to enhance transparency in neural networks.
The paper tackles the problem of interpreting neural network layers by introducing concept whitening (CW), which aligns latent space axes with known concepts, resulting in clearer understanding of concept learning across layers without compromising predictive performance.
What does a neural network encode about a concept as we traverse through the layers? Interpretability in machine learning is undoubtedly important, but the calculations of neural networks are very challenging to understand. Attempts to see inside their hidden layers can either be misleading, unusable, or rely on the latent space to possess properties that it may not have. In this work, rather than attempting to analyze a neural network posthoc, we introduce a mechanism, called concept whitening (CW), to alter a given layer of the network to allow us to better understand the computation leading up to that layer. When a concept whitening module is added to a CNN, the axes of the latent space are aligned with known concepts of interest. By experiment, we show that CW can provide us a much clearer understanding for how the network gradually learns concepts over layers. CW is an alternative to a batch normalization layer in that it normalizes, and also decorrelates (whitens) the latent space. CW can be used in any layer of the network without hurting predictive performance.