This is a *special* CWI Scientific Meeting with the two Van Wijngaarden prize laureates Marta Kwiatkowska (computer scientist ) and Susan A. Murphy (mathematician).
* 15:00 - 15:45 Marta Kwiatkowska: Safety and robustness for deep learning with provable guarantees
* 15:45 - 16:00 Questions and small break
* 16:00 - 16:45 Susan Murphy: Inference for Longitudinal Data After Adaptive Sampling
* 16:45 - end Questions followed by drinks
Marta Kwiatkowska (University of Oxford): Safety and robustness for deep learning with provable guarantees
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.
Susan Murphy (Harvard University): Inference for Longitudinal Data After Adaptive Sampling
Abstract: Adaptive sampling methods, such as reinforcement learning (RL) and bandit algorithms, are increasingly used for the real-time personalization of interventions in digital applications like mobile health and education. As a result, there is a need to be able to use the resulting adaptively collected user data to address a variety of inferential questions, including questions about time-varying causal effects. However, current methods for statistical inference on such data (a) make strong assumptions regarding the environment dynamics, e.g., assume the longitudinal data follows a Markovian process, or (b) require data to be collected with one adaptive sampling algorithm per user, which excludes algorithms that learn to select actions using data collected from multiple users. These are major obstacles preventing the use of adaptive sampling algorithms more widely in practice. In this work, we provide theory for common Z-estimators based on adaptively sampled data. The inference is valid even when observations are non-stationary and highly dependent over time, and (b) allow the online adaptive sampling algorithm to learn using the data of all users. Furthermore, our inference method is robust to miss-specification of the reward models used by the adaptive sampling algorithm. This work is motivated by our work in designing the Oralytics oral health clinical trial in which an RL adaptive sampling algorithm will be used to select treatments, yet valid statistical inference is essential for conducting primary data analyses after the trial is over.
