How Deep Neural Techniques have revolutionised Information Retrieval and Text Mining

Speaker: Mark Carman, Politecnico di Milano

Abstract:
Over the last few years, deep neural architectures have rewritten the rulebook in terms of the performance that can be achieved across a multitude of text processing tasks from sentiment analysis and sarcasm detection, to machine translation, web search, question answering, and dialog generation. In this talk I will explain the language modelling technology behind these advances, discussing its evolution from shallow embeddings to modern Transformer models composed of ever deeper self-attention networks. I will describe numerous applications of these deep models in information retrieval and text mining and then look to the future, to applications that seamlessly combine information across text and image modalities.

Biography:
Mark Carman is an Associate Professor at the Politecnico di Milano. Originally from Adelaide, Australia, he moved to Milan in 2018 after spending eight years as a Lecturer and later Senior Lecturer at Monash University in Melbourne. Before living in Melbourne, he was a postdoc at the Università della Svizzera italiana (USI) in Switzerland. He received his PhD in 2006 from the University of Trento in Italy after working at both the Fondazione Bruno Kessler (FBK) in Trento and the Information Sciences Institute (ISI) in Los Angeles, California. Mark's research lies in Data Science with a focus on Deep Learning applications in Information Retrieval, Natural Language Processing, Digital Forensics and Bioinformatics. In the past he has worked on learning of search rankings, scaling up machine learning, robust clustering, quality-control in crowd-sourcing, personalised search and recommendation, sentiment and sarcasm detection, OCR systems, and user expertise estimation, among other topics. He has authored publications at many prestigious venues, including the conferences: SIGIR, KDD, IJCAI, WSDM, WWW, CIKM, ECML-PKDD, ECIR, HT, CoNLL, EACL, HCOMP, PAKDD, ICDM, ICDAR, and the journals: IS, TOIS, IR, JMLR, ML, PR, JAIR, AI, CS&L, JASIST, DI, CSUR. He has also served on the program committees for conferences: SIGIR, WSDM, CIKM, ECIR, KDD, WWW, EMNLP, ACML, IJCAI and AAAI, and as an Associate Editor for the journal ACM TOIS.

Host: Prof. Cesare Alippi