Learning as the Unsupervised Alignment of Conceptual Systems
This addresses the problem of scaling supervised learning for concept induction, offering a novel unsupervised approach that could benefit AI systems in domains like multimodal learning, though it appears incremental in building on existing ideas from philosophy and psychology.
The paper tackles the challenge of concept induction without explicit supervision by demonstrating that conceptual systems can be built and aligned using environmental information, making learning easier as more concepts are involved. The result is supported by computational experiments showing that unique signatures across systems (e.g., images and text) enable this alignment, as observed in children's early concept formation.
Concept induction requires the extraction and naming of concepts from noisy perceptual experience. For supervised approaches, as the number of concepts grows, so does the number of required training examples. Philosophers, psychologists, and computer scientists, have long recognized that children can learn to label objects without being explicitly taught. In a series of computational experiments, we highlight how information in the environment can be used to build and align conceptual systems. Unlike supervised learning, the learning problem becomes easier the more concepts and systems there are to master. The key insight is that each concept has a unique signature within one conceptual system (e.g., images) that is recapitulated in other systems (e.g., text or audio). As predicted, children's early concepts form readily aligned systems.