‘Simplicity bias’ in science

22 May 2018

A team led by Professor Ard Louis from Oxford’s Department of Physics has shown how to make the coding theorem from algorithmic information theory practically applicable, and use it to show that many phenomena -- ranging from biological RNA structures to financial models to deep learning with neural networks -- exhibit ‘simplicity bias’, which makes simple outputs far more likely than complex outputs.

Find out more on the Oxford Science Blog.