Seminer - Burcu YILMAZ
Recent Trends in the Use of Graph Neural Network Models
Date and Time: October 30th, 2019 (Wednesday), 13:30
Place: Room Z23, Computer Engineering Building, GTU
All interested are cordially invited.
ABSTRACT:
Graphs are powerful data structures that allow us to represent varying kinds of relationships within data. Although graph structure can model the complex nature of the data, several machine learning solutions are not designed to model data as graphs. This causes an inevitable significant information loss during machine learning process. On the other hand, since structure of graphs are changeable, it is a challenging task to extract fixed size features from graphs for machine learning tasks. In the past, due to the difficulties related to the time complexities of processing graph models, graphs rarely involved in machine learning tasks. In the recent years, especially with the new advances in deep learning techniques, increasing number of graph models related to the feature engineering and machine learning are proposed.
Graph representation learning has been proved to be extremely useful for a broad range of graph-based analysis and prediction tasks. Recently, there has been an increase in approaches that automatically learn to encode graph structure (graph, subgraph or node etc.) into low dimensional embedding (vector representation). These approaches are accompanied by models for machine learning tasks, and they fall into two number of categories. The first one focuses on feature engineering techniques on graphs and reformats input graphs to fit the state of the art machine learning techniques. The second group of models (GCN models, graph2sec etc.) assembles graph structure to learn a graph neighbourhood representation in the machine learning model itself.
Graphs in machine learning have many application areas such as chemistry and social network analysis. In the seminar, after giving a few general examples in varying applications, the advances in applications of graphs on natural language processing using the recent deep learning models will be focused.
BIOGRAPHY:
Burcu YILMAZ is an assistant professor in Institute of Information Technologies at the Gebze Technical University. She received her PhD degree in computer science from the Gebze Institute of Technology in 2010. Burcu is conducting research in data mining, machine learning, natural language processing, deep learning, graph mining, graph neural networks and social network analysis.
