Reasoning-Modulated Representations
This work addresses the challenge of data efficiency in representation learning for AI systems, though it appears incremental as it builds on existing self-supervised methods.
The paper tackles the problem of neural networks requiring large training sets to learn robust internal representations by incorporating prior knowledge about the underlying system into a pre-trained reasoning module, showing that this approach improves representation learning in self-supervised settings from pixels.
Neural networks leverage robust internal representations in order to generalise. Learning them is difficult, and often requires a large training set that covers the data distribution densely. We study a common setting where our task is not purely opaque. Indeed, very often we may have access to information about the underlying system (e.g. that observations must obey certain laws of physics) that any "tabula rasa" neural network would need to re-learn from scratch, penalising performance. We incorporate this information into a pre-trained reasoning module, and investigate its role in shaping the discovered representations in diverse self-supervised learning settings from pixels. Our approach paves the way for a new class of representation learning, grounded in algorithmic priors.