DreamCoder: Growing generalizable, interpretable knowledge with wake-sleep Bayesian program learning
This addresses the challenge of building generalizable, interpretable knowledge systems for AI, though it appears incremental in combining existing ideas like program synthesis and wake-sleep learning.
DreamCoder tackles the problem of acquiring expertise by learning programming languages and concepts through a wake-sleep Bayesian program learning approach, resulting in the system solving inductive programming tasks and creative tasks like drawing, while rediscovering fundamental concepts like functional programming and Newton's laws.
Expert problem-solving is driven by powerful languages for thinking about problems and their solutions. Acquiring expertise means learning these languages -- systems of concepts, alongside the skills to use them. We present DreamCoder, a system that learns to solve problems by writing programs. It builds expertise by creating programming languages for expressing domain concepts, together with neural networks to guide the search for programs within these languages. A ``wake-sleep'' learning algorithm alternately extends the language with new symbolic abstractions and trains the neural network on imagined and replayed problems. DreamCoder solves both classic inductive programming tasks and creative tasks such as drawing pictures and building scenes. It rediscovers the basics of modern functional programming, vector algebra and classical physics, including Newton's and Coulomb's laws. Concepts are built compositionally from those learned earlier, yielding multi-layered symbolic representations that are interpretable and transferrable to new tasks, while still growing scalably and flexibly with experience.