We are very happy to announce the two great invited talks to take place at TextGraphs-12.
Maximilian Nickel: Hierarchical Representation Learning on Graphs Abstract: Many domains such as natural language understanding, information networks, bioinformatics, and the Web are characterized by problems involving complex graph-structures and large amounts of uncertainty. Representation learning has become an invaluable approach for making statistical inferences in this setting by allowing us to learn high-quality models on a large scale. However, while complex graphs often exhibit latent hierarchical structures, current embedding methods do not account for this property. This leads not only to very inefficient representations but also to a reduced interpretability of the embeddings.
In this talk, I will first give a brief overview over state-of-the-art methods for learning representations of relational data such as graphs and text. I will then introduce a novel approach for learning hierarchical representations by embedding relations into hyperbolic space. I will discuss how the underlying hyperbolic geometry allows us to learn parsimonious representations which simultaneously capture hierarchy and similarity. Furthermore, I will show that hyperbolic embeddings can outperform Euclidean embeddings significantly on data with latent hierarchies, both in terms of representation capacity and in terms of generalization ability.
Short Bio: Maximilian Nickel is a postdoctoral fellow with the Laboratory for Computational and Statistical Learning (LCSL), hosted at the Center for Biological and Computational Learning (CBCL), Massachusetts Institute of Technology (MIT). In 2013, he received the Ph.D. summa cum laude from the Department for Informatics of the Ludwig-Maximilians-University of Munich under supervision of Volker Tresp. He received a diploma degree with honors in computer science from the Ludwig Maximilian University Munich in 2009. The main focus areas of his research are: Machine Learning from Structured Data and Knowlede Graphs, Representation Learning and Structured Embeddings, Statistical Relational Learning, Knowledge Representation, Tensor Factorizations & Multilinear Algebra.
Gabor Melli: Neural Domain-Specific Wikification Abstract: Wikification is the process of annotating text with a wikitext markup language, such that relevant spans of text are recognized and linked to their corresponding entry in a wiki. This talk explores the application of recent neural-based NLP advances to the task with a focus on domain-specific wikis.
Short Bio: Gabor is the Senior Director of ML & AI at PlayStation Network. He has twenty-plus years of experience in the delivery of automated data-driven solutions in a technical lead and management capacity. His passion is to introduce semantic capabilities into mission critical processes. He has led over twelve large-scale data-driven initiatives at both enterprises ranging from Sony PlayStation, Microsoft, AT&T, T-Mobile, ICBC, Washington Mutual and Wal*Mart, and at start-ups such as Datasage (acquired by Vignette/Open Text), Meals.com, PredictionWorks, VigLink, and OpenGov. He continues to publish, present and organize world-class conferences that focus on applied dynamic semi-supervised semantic automation.