Haris Papageorgiou’s topics:
1.Neural Topic Models
Student(s) will be acquainted with and capable of exploring recent advances in data mining. Concretely, the focus here is on the applicability of deep learning in Topic Modeling on large textual collections. The data collections, the core of the methodologies to be explored and the interactive tools will be available at the initial phase of the work.
2. Question Answering and Knowledge Graphs
The aim of this work is to analyze and isolate interesting insights in scientific collections. At a next step, we will be using these elements to populate a knowledge base enabling us to pose complex questions. We plan to make progress beyond document and snippet extraction which are typical goals in the mainstream question answering systems. The data collections, the core of the methodologies to be explored and the interactive tools will be available at the initial phase of the work.
3.Conversational Agents and Deep Learning
We will investigate the modeling of dialogue flows with deep neural techniques. The work will be based on mature platforms enabling student(s) to explore and cope with the critical areas in dialogue modeling. The application domain as also as the feasiblity study will be part of the preliminary stages of the work.
4.AI in Industry
We will exploit state-of-the-art ML/DL techniques in other disciplines. Among the candidates to be explored are interesting application areas in NanoMetrology, Medicine, Fisheries and aquaculture, etc.