DEAR COLLEAGUES!
We invite you to participate in the scientific seminar "Theory of Optimal Solutions ", which will take place on March 31, 2026, at 16:00.
Speaker: Kozyriev Anton Yuriiovych, Ph.D. student (National Technical University of Ukraine "Igor Sikorsky Kyiv Polytechnic Institute")
Topic: Stochastic quantization methods for scalable cluster analysis of big data
Abstract: Common centroid-based clustering methods for indexing large multimedia data in the form of latent vector features are considered. Data indexing allows one to speed up the search for relevant to queries data in large databases. The optimal clustering problem is interpreted as a nonlinear transportation problem with movable consumption centers and is reduced to a non-smooth non-convex stochastic optimization problem. An original adaptive clustering method based on adaptive stochastic approximation technique is proposed to reduce the required iterations to guarantee convergence and the possibility of using non-quadratic objective functions that improves the robustness to statistical noise. Examples of the method application are given.
Seminar Leader: Corresponding Member of the NAS of Ukraine Stetsyuk P.I.
https://us04web.zoom.us/j/5377511780?pwd=V1FYSmhBNjlFR2RNRWxCdWVlQUUwZz09
Conference ID: 537 751 1780
Access code: 3HHTxH
Everyone interested can join and participate in scientific discussions.