Alexey Zaytsev
Skoltech, BIMSA
20th July 2023, 04:00pm - 05:00pm (GST)
| Title: | Deep models for processing event sequences |
| Abstract: | Massive samples of event sequences data occur in various domains, including e-commerce, healthcare, and finance. There are two main challenges regarding the inference of such data: methodological and computational. The methodological peculiarity for event sequences is their non-uniformity and sparsity, making time series models unsuitable. The computational challenge arises from a large amount of available data and the significant length of each sequence. Thus, it requires complex models. In the talk, I will review existing solutions include large recurrent and transformer neural network architectures, and how they adapt to the setting of event sequences modeling as temporal point processes. In particular, we consider the self-excitement effect and continuous time, and how they can be handled within existing approaches. Additional attention I will pay to the self-supervised learning within the event sequences framework. |
| Bio: | Alexey is a well-known scientist in the area of deep learning and machine learning. His research focuses on developing new techniques for sequential data and enhancing the robustness of deep learning models, specifically within the realms of probabilistic machine learning and representation learning for diverse data modalities. In this areas, he obtained results presented at major conference like ACM MM, KDD, and AISTATS. He also pay attention to industry by providing innovative solution that already are in production in major banks and oil&gas companies. |