Here I summarise some results from my thesis which are relevant to artificial intelligence (AI) and machine learning (ML). The key contribution is a theory of optimally sample and energy efficient learning, which is supported by mathematical proofs and experimental results.