Skip to main content

Renowned data mining expert Professor Kyuseok Shim unleashes the power of deep learning in cardinality estimation

Photo 1
Professor Kyuseok Shim presents his groundbreaking research on deep learning for cardinality estimation in approximate substring queries, paving the way for transformative advancement in database systems.

Professor Kyuseok Shim, a renowned data mining expert from Seoul National University in Seoul, South Korea, delivered a highly insightful lecture titled “Cardinality Estimation of Approximate Substring Queries using Deep Learning” on July 18, 2023 (Tuesday). His research focused on revolutionising cardinality estimation for approximate substring queries in database systems by harnessing the power of deep learning.


Traditionally, cardinality estimation relied on statistical assumptions and summaries, but the remarkable success of deep learning models in capturing complex data patterns for general queries inspired Professor Shim to explore the model’s potential for approximate substring queries. To achieve this, Professor Shim and his research team developed efficient algorithms to generate training data, significantly reducing computational costs for cardinality estimation. They achieved swift and accurate estimations by introducing a custom deep-learning model and a novel learning method. Extensive experiments demonstrated the superiority of the deep learning approach over traditional methods, offering promising prospects for substantial advancements in database system performance.


The lecture concluded with a fruitful and engaging Q&A session, where participants eagerly explored the intricacies of the packed-learning method. Many insightful questions were raised, particularly concerning the potential applications of advanced feature extraction techniques across various domains, sparking lively discussions.


Before joining Seoul National University, Professor Shim was an Assistant Professor at the Korea Advanced Institute of Science & Technology (KAIST). His exceptional research contributions span various areas, including data mining, machine learning, privacy preservation, query processing, query optimization, data warehousing, semi-structured data (XML), stream data, and histograms. His expertise in data mining and query processing has earned him prestigious titles of the fellow of Association for Computing Machinery (ACM) and the Institute of Electrical and Electronics Engineers (IEEE).


Please click here for the full video of the Lecture and here for more photos.