Prof Vazirgiannis is a visiting professor supported by the "Wallenberg AI, Autonomous Systems and Software Program" (WASP). His areas of research is machine and deep learning for large scale heterogeneous data (including graphs and text). More specifically in the last years we have been researching community detection, graph classification/clustering/embeddings, influence maximisation, graph-based NLP, deep learning for word embeddings with applications to digital marketing, event detection, summarisation and legal text.
Machine and recently deep learning are significant methods to help extracting knowledge from large scale data in all domains. The most interesting and expressive data are graphs that are present in many domains including biology, social networks, power/communication/transport facilities. In this domain the topic that really fascinates me is deep learning on graphs, known as Graph Neural networks. This area is really booming while state of the art is advancing rapidly with promising potential for many real-life applications. Overall AI promises to change our lives significantly and we need to be well prepared to maximise the benefits to society and economy in a balanced manner
The methods we are developing are very relevant to different application domains where text and graphs are present. Deep learning in the presence of large-scale data enables extracting the best possible models for different predictive and unsupervised tasks for different domains including bio/medicine, legal documents, social networks, online marketing and advertising, recommendations, retail etc. In the future, many processes in the above domains will be fully automated and mastering algorithms in this case is important.