LMS Invited Lectures on the Mathematics of Deep Learning

Neural networks were originally introduced in 1943 by McCulloch and Pitts as an approach to develop learning algorithms by mimicking the human brain. The key goal at that time was the introduction of a theory of artificial intelligence. However, the limited amount of data and the lack of high performance computers made the training of deep neural networks, i.e., networks with many layers, unfeasible.

28 Feb 2022 from 2 p.m. to 4 Mar 2022 3 p.m. CET (GMT+0100)

Dear colleagues,

Thanks for joining our CWI event where we watch the “Lectures on the Mathematics of Deep Learning” (LMS) workshop together. I am looking forward to discussing the talks with you all. The LMS workshop will be from Monday 28th of February untill Friday the 4th of March. We have rooms available with a beamer on all days. The schedule at the CWI will be as follows:

28-02: room L017, time 14:00 – 17:00
        14:00 – 15:15   Introduction to Deep Neural Networks
        16:00 – 17:00   Deep Neural Networks: From Approximation to Expressivity

01-03: room L017, time 10:00 – 11:00  +  14:00 – 15:00
        10:00 – 11:00   Deep Neural Networks: Analysing the Training Algorithm
        → Physical event we cannot join
        14:00 – 15:00   Deep Neural Networks: The Mystery of Generalisation
        15:30 – …   Optional discussion via Gather Town

02-03: room L120, time 10:00 – 17:00
        10:00 – 11:00   Deep Neural Networks: Opening the Black Box via Explainability Methods
        11:30 – 12:30   Deep Neural Networks: Towards Robustness
                Lunch break
        14:00 – 15:30   Posters and Networking At INI and in Gathertown
        16:00 – 17:00   TBC

03-03: room L017, time 10:00 – 16:30
        10:00 – 11:00   Limitations of Deep Neural Networks
        11:30 – 12:30   Machine Learning for the Sciences: Towards Understanding
                Lunch break
        14:00 – 15:00   Inverse Problems meet Deep Learning: Optimal Hybrid Methods
        15:30 – 16:30   Benign Overfitting in Linear and Nonlinear Settings

04-03: room L120, time 10:00 – 15:15
        10:00 – 11:00   TBC
        11:30 – 12:30   Partial Differential Equations meet Deep Learning: Beating the Curse of Dimensionality
                Lunch break
        14:00 – 15:00   Mathematical Foundations of Deep Learning: Potential, Limitations, and Future Directions
        Closing remarks

PLEASE NOTE: It seems unlikely that people will be able to join our CWI event virtually given how LMS will be organized. This means that you will probably have to join physically at the CWI for those talks you want to follow. This may be somewhat difficult with the work restrictions still in place (we don’t know what the situation will be during LMS). Therefore, please try to arrange physical attendance for those lectures you definitely want to see. If this is somehow difficult given office spaces, please let me know.

Since the programme is quite extensive, you may not want to follow every lecture. However, please try to join as many as possible to enable interesting discussion in the group afterwards.  The full programme can also be found here and some more information can be found here.