CLS Speaker Series: Ivan Titov - Today, Tuesday (13/02), 16:00, F1.15, SP 107

[Elia Bruni]

Hi all,

I am happy to announce that our next invited speaker is Ivan TItov (Universities of Edinburgh (ILCC) and Amsterdam (ILLC)) today, Tuesday, 13/02, at 16:00 in room F1.15.

Best,

Elia

---

Graph Convolutional Networks for Natural Language Processing and Relational Modeling

Graph Convolutional Networks (GCNs) is an effective tool for modeling graph structured data. We investigate their applicability in the context of natural language processing (machine translation and semantic role labelling) and modeling relational data (link prediction). For natural language processing,  we introduce a version of GCNs suited to modeling syntactic and/or semantic dependency graphs and use them to construct linguistically-informed sentence encoders. We demonstrate that using them results in a substantial boost in machine translation performance and state-of-the-art results on semantic role labeling of English and Chinese. For link prediction, we propose Relational GCNs (RGCNs), GCNs developed specifically to deal with highly multi-relational data, characteristic of realistic knowledge bases. By explicitly modeling neighbourhoods of entities, RGCNs accumulate evidence over multiple inference steps in relational graphs and yield competitive results on standard link prediction benchmarks. 

Joint work with Diego Marcheggiani, Michael Schlichtkrull, Joost Bastings, Thomas Kipf, Khalil Sima’an, Max Welling,  Rianna van den Berg and Peter Bloem.