Keynote Speakers

The following speakers have graciously accepted to give keynotes at AIST-2024.

Alexey Zaytsev

Alexey Zaytsev is the Head of the Laboratory of Applied Research for Structured Data Statistics and an Assistant Professor at Skoltech. He graduated from Phystech and earned his PhD in Math and Physics in 2017 from IITP RAS. Alexey is also the author of numerous publications at various venues including A*-rank conferences. His research interests include neural modeling of event sequences and uncertainty estimation in deep learning for various data modalities.

Self-Supervised Learning for Event Sequences

Abstract: Event sequences are a crucial yet often overlooked data modality, despite their significant industrial importance. This talk will highlight our advancements in this field, that benefited both academia and industry in recent years.

To set the stage, we will explore two approaches to modeling event sequences and examine their representation learning capabilities. On the theoretical side, we will delve into the generalization of inhomogeneous Poisson processes, known as the Hawkes process, as a model for event sequences. On the practical side, we will demonstrate how natural language processing (NLP) techniques can be adapted to this domain.

Our results show that the derived representations are highly effective in various applications, including credit scoring and churn detection. The talk will conclude with a discussion on the robustness of these models, addressing common challenges and presenting solutions to enhance their reliability.

Anton M. Alekseev & Timur Turatali

Anton M. Alekseev currently holds research and teaching roles at Kazan (Volga Region) Federal University (KFU), Steklov Mathematical Institute in St. Petersburg (PDMI RAS), and Saint Petersburg State University (Mathematics and Computer Science Faculty). He is also pursuing a PhD at the Kyrgyz State Technical University named after I. Razzakov. Anton has more than 10 years of experience in industry and academia. His industrial experience includes high-load service development, real-world data fuzzy matching, chatbot design, data annotation management, and machine learning systems engineering. In academia, Anton conducts both basic research and research projects for various commercial organizations at KFU and PDMI RAS. His interests include natural language processing, deep learning, and machine learning for software engineering (ML4SE). Anton received the Best NLP Paper Award at AIST-2023.

Timur Turatali is a Machine Learning Engineer and Data Scientist with over 8 years of experience, having made significant contributions in both startup environments and established companies like EY and Citibank. Beyond his corporate career, Timur co-founded the AI community of Kyrgyzstan, which now boasts over 1,600 members, including Machine Learning engineers, AI specialists, and Data Scientists. He is also an active contributor to open-source Natural Language Processing (NLP) projects for the Kyrgyz language, working on initiatives like Named Entity Recognition (NER), spell checking, and large language models (LLMs). In support of AI development for the Kyrgyz language, he co-founded an open-source initiative called AkylAI and The Cramer Project. AkylAI is the first AI chatbot and smart speaker in Kyrgyz. Timur has played a key role in fostering the tech ecosystem by organizing more than 10 hackathons, over 50 meetups, and a major networking event focused on IT, AI, and Machine Learning. His research interests include Natural Language Processing for less resourced languages, as well as deep learning and machine learning applications for mental health.

KyrgyzNLP: Challenges, Progress, and Future

Abstract:The large language models (LLMs) have excelled in numerous benchmarks, advancing AI applications in both linguistic and non-linguistic tasks. However, this has primarily benefited well-resourced languages, leaving less-resourced ones (LRLs) at a disadvantage. In this talk, we will highlight the current state of the NLP field in the specific LRL: кыргыз тили.

Human evaluation, including annotated datasets created by native speakers, remains an irreplaceable component of reliable NLP performance, especially for LRLs where automatic evaluations can fall short. In assessments of the resources for Turkic languages, Kyrgyz is labeled with the status ‘Scraping By’, a severely under-resourced language spoken by millions. This is concerning given the growing importance of the language, not only in Kyrgyzstan but also among diaspora communities where it holds no official status.

We review prior efforts in the field, noting that many of the publicly available resources have only recently been developed, with few exceptions beyond dictionaries. While recent papers have made some headway, much more remains to be done. Despite interest and support from both business and government sectors, the situation for Kyrgyz language resources remains challenging. We stress the importance of community-driven efforts to build the latter, ensuring the future advancement of sustainability. We then share our view of the most pressing challenges in Kyrgyz NLP. Finally, we propose a roadmap for future development in terms of research topics and language resources.