BaCaTeC. Control and Machine Learning

BaCaTeC. Control and Machine Learning

  • Enrique Zuazua, Miroslav Krstic
  • Supported by HighTech Research between Bavaria and California (BaCaTeC, Bayerisch-Kalifornische Hochschulzentrum)
  • Duration: 2023 – 2025
  • Machine Learning (ML) is forging a new era in Applied Mathematics (AM), generating rich, intensive, innovative research, insightful new ideas and powerful methods. However, simultaneously, this is leading to very challenging fundamental, deep and complex mathematical questions since ML needs reliable methods with theoretical guarantees that can explainably assure good performance and generalisation properties. This also has important ethical consequences and implications.

    This great challenge can be addressed through connection with other, more mature areas of AM, from multiple perspectives. This project seeks to develop the rich interface between Control Theory (CT) and ML. We also aim to contribute to building the analytical foundations of the theory of ML needed to understand and improve the computational efficiency of some methods in ML, such as those arising in the context of Residual Neural Networks (ResNets) for Supervised Learning (SL) and Federated Learning (FL), which will also significantly enlarge and update the range of applications of CT.

    People involved

      
    Read more at BaCaTec
    Read more at BaCaTec (Deutsch)