Please join us for the next Scientific Meeting on Friday 19 November (13.00-14.00).
During this meeting, the two 2021 Van Wijngaarden award winners Marta Kwiatkowska and Susan Murphy will give a lecture in the Euler room.
Safety and robustness for deep learning with provable guarantees
Marta Kwiatkowska (University of Oxford)
Abstract: Computing systems are becoming ever more complex, with decisions increasingly often based on deep learning components. This lecture will describe progress with developing automated verification techniques for deep neural networks to ensure safety and robustness of their decisions. The lecture will conclude with an overview of the challenges in this field.
Causal Inference Using Adaptively Collected Data
Susan Murphy (Harvard University)
Abstract: Often data from multiple individuals is independent. However data collected by use of online algorithms is " adaptively collected." In this case of sequential treatments (e.g. website ads, medications, behavioral treatments), the data collected prior to time t, determines which treatment is delivered and hence which data is collected at time t. Quite interestingly, the variance of common estimators such as a treatment effect may not converge even in the limit as the number of individuals increases. This talk concerns how one can, none-the-less, assess treatment effects in these settings.