Research Class: Text Representation Models Based on Neural Networks

Datum održavanja: četvrtak, 6.2.2020. u 12:00 sati, prostorija O-358
Predavač: Karlo Babić,  Odjel za informatiku
Naziv predavanja: Text Representation Models Based on Neural Networks

Abstract:

Text is an important source of human knowledge and a medium through which humans communicate. For artificial intelligence in general, it is very useful to give computers the ability to understand language. To do so, text first has to be transformed into text representation (represented as vectors of numbers). It is desirable that text representations are encoded in a vector space that preserves information about semantics, syntax, knowledge, and other information that could be useful for downstream tasks (tasks that use text representation). Such a representation would be closer to the human way of perceiving text. Text representation is directly needed in tasks such as information retrieval, question answering, textual entailment identification, keyword extraction, text summarization, sentiment analysis, machine translation, etc. This talk will give an overwiev of a large number of neural network models that learn text representations.

Podijeli članak