Model reduction across disciplines and approaches
Seminar by Max Tschaikowski, Associate Professor in Computer Science, Sapienza, Rome. Talk on Model reduction, with applications to Quantum computing, Machine Learning, Network science, and Biology (among others).
In association with the L'EMbeDS Data Science Seminar Series
Abstract
Model reductions compress an original model to a smaller one while preserving properties of interest, thus allowing to reduce simulation and computation times. Network embedding, for instance, allows to project networks onto lower-dimensional spaces while preserving similarities among nodes. Similarly, quantum bisimulation compresses quantum circuits without altering a quantum measurement of interest, while differential equivalence achieves the same for chemical reaction networks in case of biological measurement. In this talk, we will give a high-level overview of model reductions that arise from linear invariant subspaces. The first part will focus on graph-based, while the second part will discuss geometry-based approaches. Their applicability will be demonstrated on benchmarks from network science, systems biology and quantum computing.
- When: Friday, April 17 | 10:00 AM
- Where: Aula 2, Main Campus
- Online: Join via Webex
Speaker
Max Tschaikowski, Sapienza University of Rome