In May 2020, the Senate of the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) established the Priority Programme “Theoretical Foundations of Deep Learning” (SPP 2298). The programme is designed to run for six years. The present call invites proposals for the first three-year funding period.

We currently witness the impressive success of deep learning in real-world applications, ranging from autonomous driving over game intelligence to the health care sector. At the same time, deep learning-based methods have a similarly strong impact on science, often replacing state-of-the-art classical model-based methods to solve mathematical problems such as inverse problems or partial differential equations. However, despite this outstanding success, most of the research on deep neural networks is empirically driven and mathematical foundations are largely missing. Moreover, in several special but important cases these techniques dramatically fail under small perturbations such as adversarial examples in image classification, which calls for improvements driven by a theoretical underpinning.

The key goal of this Priority Programme is the development of a comprehensive theoretical foundation of deep learning. The research within the programme will be structured along three complementary points of view, namely

1. the statistical point of view regarding neural network training as a statistical learning problem and studying expressivity, learning, optimisation, and generalisation,
2. the applications point of view focusing on safety, robustness, interpretability, and fairness, and
3. the mathematical methodologies point of view developing and theoretically analysing novel deep learning-based approaches to solve inverse problems and partial differential equations.

The research questions to be addressed within this Priority Programme are of a truly interdisciplinary nature and can only be solved by a joint effort of mathematics and computer science. Mathematical methods and models throughout mathematics are required, including algebraic geometry, analysis, applied probability, approximation theory, differential geometry, discrete mathematics, functional analysis, optimal control, optimisation, and topology. A fundamental role is similarly played by statistics as well as theoretical computer science. In this sense, methods from mathematics, statistics, and computer science are at the core of this Priority Programme.

Successful proposals address a genuine contribution to the understanding and the theoretical foundations of deep learning along the above three complementary points of view. Projects aiming “only” at the application of existing methods of deep learning or their further development without theoretical foundations may not be funded in the frame of the programme.

Proposals must be written in English and submitted to the DFG by 30 November 2020. Please note that proposals can only be submitted via “elan”, the DFG’s electronic proposal processing system. To enter a new proposal within the existing Priority Programme, go to Proposal Submission – New Project/Draft Proposal – Priority Programmes and select “SPP 2298” from the current list of calls.

Applicants must be registered in “elan” prior to submitting a proposal to the DFG. If you have never before submitted a proposal to DFG through “elan”, you need to register in advance. This can be done online by yourself – however, it takes one to two working days to be confirmed by DFG staff. If you need to register, please complete your registration before 16 November 2020. Note that you will be asked to select the appropriate Priority Programme call during both the registration and the proposal process. If your contact data in “elan” is outdated, please also update it before that date.

General information on proposals in the framework of a Priority Programme (in particular concerning eligibility and admissible funding requests) can be found in guideline 50.05 (part B). See also guideline 54.01 for instructions how to prepare a proposal. The specific proposal has to be structured according to form 53.01. However, it is admissible to prepare the proposal as a pdf-file, e.g., using LaTeX, instead of using the rtf-file which is available online.

Note that the descriptions of the projects and all CVs need to be prepared in English. Further, DFG’s rules for publication lists (guideline 1.91) need to be respected: Beside the general bibliography, every proposal should include a list of up to ten publications by the applicant(s) (and/or members of her/his/their group) that relate directly to the project. Any academic CV submitted to the DFG must not list more than ten publications, which describe best the scientist’s profile. Publications in these lists need to be classified as a) refereed publications (published articles and monographs; accepted articles with note of acceptance by the journal) or b) other publications (e. g., preprints on arXiv), which need to be accessible online (please mention the arXiv number or provide links).

With the submission of a proposal within this programme, the applicants agree that DFG shares all necessary information with the coordinator of the Priority Programme after the call deadline.

Due to the Coronavirus-pandemic the review will not take place in form of a colloquium. It is planned to have a video review meeting by a confidential international review panel in Spring 2021. It is expected to submit additional posters for the suggested projects in due time before the virtual panel meeting that e.g. may also contain updated information regarding latest publications. Information about the poster format and submission date will be given later to the applicants by the coordinator of the Priority Programme. The decision about the projects will be reached in the summer of 2021.

Further Information

The elan system can be accessed at:

• Externer Linkhttps://elan.dfg.de/en 

DFG forms 50.05 and 54.01 can be downloaded at:

• Interner Linkwww.dfg.de/formulare/50_05 
• Interner Linkwww.dfg.de/formulare/54_01 

For scientific enquiries please contact the Priority Programme’s coordinator:

• Prof. Dr. Gitta Kutyniok,
  Technische Universität Berlin,
  Fakultät II – Mathematik und Naturwissenschaften,
  Institut für Mathematik,
  Straße des 17. Juni 136,
  10623 Berlin,
  phone +49 30 314-25758,
  Link auf E-Mailkutyniok@math.tu-berlin.de 

Questions on the DFG proposal process can be directed to:
Programme contact:

• Dr. Frank Kiefer,
  phone +49 228 885-2567,
  Link auf E-Mailfrank.kiefer@dfg.de

Administrative contact:

• Silke Seiler,
  phone +49 228 885-2751,
  Link auf E-Mailsilke.seiler@dfg.de

Note:

This text is available at Externer Linkwww.dfg.de/foerderung/info_wissenschaft/2020/info_wissenschaft_20_36 
Please use this identifier to cite or link to this item.