Please use this identifier to cite or link to this item:
|Title:||Graph neural networks for questions and answers||Authors:||Mah, Caleb||Keywords:||DRNTU::Engineering::Computer science and engineering::Computing methodologies::Artificial intelligence
DRNTU::Engineering::Computer science and engineering::Computing methodologies::Document and text processing
|Issue Date:||2019||Abstract:||The role of machine learning algorithms in natural language processing (NLP) tasks has become increasingly important. In the pursuit of developing intelligent agents capable of not only understanding but also reasoning about natural language, it can be beneficial to formulate such problems as graphs for which recent neural network techniques are able to interpret. One example of such a task would be the ability to answer queries about a given set of statements. This project explores the effectiveness of graph structured deep learning techniques in generating answers to questions. In particular, the relative performance of a new technique involving residual gated convolutional networks will be compared against earlier methods using the bAbi tasks dataset as a benchmark. We will show that this new model performs similarly, if not better than, existing models and discuss the limitations faced.||URI:||http://hdl.handle.net/10356/76956||Rights:||Nanyang Technological University||Fulltext Permission:||restricted||Fulltext Availability:||With Fulltext|
|Appears in Collections:||SCSE Student Reports (FYP/IA/PA/PI)|
Items in DR-NTU are protected by copyright, with all rights reserved, unless otherwise indicated.