Structured (De)composable Representations Trained with Neural Networks
This addresses the need for interpretable and modular representations in machine learning, though it appears incremental as it builds on existing representation learning methods.
The paper tackles the problem of learning structured and composable representations for concept classes from images and labels, using deep learning to decompose representations into class and environment factors, and demonstrates results on classification and retrieval tasks across visual and language modalities.
The paper proposes a novel technique for representing templates and instances of concept classes. A template representation refers to the generic representation that captures the characteristics of an entire class. The proposed technique uses end-to-end deep learning to learn structured and composable representations from input images and discrete labels. The obtained representations are based on distance estimates between the distributions given by the class label and those given by contextual information, which are modeled as environments. We prove that the representations have a clear structure allowing to decompose the representation into factors that represent classes and environments. We evaluate our novel technique on classification and retrieval tasks involving different modalities (visual and language data).