Molecules are everywhere. Sometimes you need to figure out what a specific molecule is. There are numerous approaches to infer the molecule structure. One of them deals with a sequential procedure which is observed at every step with a lot of noise. The beauty is that this approach is cheap and scalable.

The approach involves inference on the unobserved part of the experiment which generates the noisy observation. After understanding the latent process model Bayesian inference allowed to overcome influence of noise. In the project I developed parameter recovery study to understand which parts of the process are possible to control and which are not.