PhD Fellowship in Simulation-Supervised Machine Learning for Biology

Opportunity summary

Official occupation: ph.d., naturvidenskab og teknik. Employment conditions: Almindelige vilkår. Employer CVR: 29979812. Primary career market: Public Sector, Education & Impact (occupation evidence). Department of Computer Science (DIKU) invites applicants for PhD fellowships in Computer Science. This project is a part of the project “Differentiable Realism from AI and Modeling” (DREAM) financed by the Novo Nordisk Foundation. Start date is (expected to be) 1 January 2027 or as soon as possible thereafter. The position is for 3 years. The project This PhD project will develop simulation-supervised machine learning methods for biological data where reliable labels are scarce, expensive, or impossible to obtain. The aim is to train models on synthetic data generated by biophysical simulations by developing simulations that can be automatically tuned to be realistic enough that the trained models generalize to real experimental data. The project will combine differentiable programming, biophysical simulation, differentiable rendering or signal generation, and machine-learning architectures for reducing the simulation-to-reality gap. A central challenge is to automatically tune simulations so that synthetic datasets match the variability, noise, and structure of real, unlabelled biological datasets1 …

This is a shortened extract from the public vacancy. Follow the original posting for the complete and authoritative description.