Graph Convolutional Networks

Graph Convolutional Networks (GCN) The general idea of GCN is to apply convolution over a graph. Instead of having a 2-D array as input, GCN takes a graph as an input. Why Graphs? Many problems are graphs in true nature. In our world, we see many data are graphs, such as molecules, social networks, and paper citations networks. T... Read more 14 Aug 2021 - 4 minute read

Graph-Based Reasoning over Heterogeneous External Knowledge for Commonsense Question Answering

Task introduction Task overview Using CSQA (Common Sense Questions and Answers) dataset. For questions that do not mention background knowledge, it is required to consider background knowledge and answer. Task form Input: Question Q=q_1 q_2…q_m and candidate answer set A={a_1,a_2,…,a_n} containing n answers. Goal: select the correct answer ... Read more 11 Aug 2021 - 8 minute read

Mixed Precision Training

A High-Level Overview Deep Neural Networks (DNNs) have achieved breakthroughs in several areas, including Computer Vision, Natural Language Understanding, Speech Recognition tasks, and many others. Although increasing network size typically improves accuracy, the computational resources also increase (GPU utilization, Memory). So new techniq... Read more 28 May 2020 - 7 minute read