Наредни састанак Семинара биће одржан онлајн у среду, 10. априла 2024. са почетком у 19 часова.
Предавачи: Гоце Трајчевски, Kingland Associate Professor, Department of Electrical and Computer Engineering, Iowa State University
Наслов предавања: LINK PREDICTION IN HETEROGENEOUS EVOLVING NETWORKS
Апстракт: Link prediction aims at identifying missing links, or links that are likely to be formed at some future evolution of a given graph/network. It is at the core of many applications, including prediction of a friendship in a social network, relations between entities in a knowledge graph, affinities between users and items in a recommender system, and potential biological interactions between drugs and diseases. In particular, the Temporal Heterogeneous Networks (THNs) investigate the structural interactions and their evolution in graphs with multiple types of nodes and edges. More often than not, THNs describe the evolving networks as a sequence of graph-snapshots and adopt mechanisms from static heterogeneous networks to capture spatio-temporal correlations. However, this confines the evolution to discrete-time settings often yielding an implementation of stacked mechanisms that introduce a high level of complexity. In this talk, after an overview of the problem-domain, we will focus on STHN (Simplified THN) that aims at enabling a continuous-time links prediction. We will discuss the encodings/representation that captures the evolving interactions and present an approach based on patching techniques, followed by experimental observations for both transductive and inductive scenarios. Subsequently, we will touch upon the aspect of explainability in the context of THNs.
Напомена: Регистрациона форма за учешће на Семинару је доступна на линку:
https://miteam.mi.sanu.ac.rs/asset/CW5nJWDSEZDj7p32p
Уколико желите само да пратите предавање без могућности активног учешћа, пренос је доступан на линку:
https://miteam.mi.sanu.ac.rs/asset/4LNW8WtML7rLKojoz