Ben specializes in unsupervised and self-supervised deep learning, guided by the belief that valuable insights can be uncovered from raw, unlabelled data. He develops tools that empower life scientists to apply deep learning without the need for specially collected datasets. With a strong foundation in Bayesian statistics and information theory, Ben takes a first-principles approach to model development. He enjoys the creative problem-solving required in the life sciences, particularly in image restoration and analysis where ideal data is often scarce.