This Probably Looks Exactly Like That: An Invertible Prototypical Network
This work addresses the need for more robust and interpretable concept-based machine learning without concept annotations, representing an incremental improvement in explainable AI.
The paper tackles the human-machine semantic gap in prototypical neural networks by proposing an invertible, generative approach called ProtoFlow, which sets a new state-of-the-art in joint generative and predictive modeling and achieves comparable predictive performance to existing methods while enabling richer interpretation.
We combine concept-based neural networks with generative, flow-based classifiers into a novel, intrinsically explainable, exactly invertible approach to supervised learning. Prototypical neural networks, a type of concept-based neural network, represent an exciting way forward in realizing human-comprehensible machine learning without concept annotations, but a human-machine semantic gap continues to haunt current approaches. We find that reliance on indirect interpretation functions for prototypical explanations imposes a severe limit on prototypes' informative power. From this, we posit that invertibly learning prototypes as distributions over the latent space provides more robust, expressive, and interpretable modeling. We propose one such model, called ProtoFlow, by composing a normalizing flow with Gaussian mixture models. ProtoFlow (1) sets a new state-of-the-art in joint generative and predictive modeling and (2) achieves predictive performance comparable to existing prototypical neural networks while enabling richer interpretation.