The 2017 ASC Student Supercomputer Challenge (ASC17) has announced that the contest will include an AI traffic prediction application provided by the Baidu Institute of Deep Learning. Commonly used among unmanned vehicle technologies, this key application software assesses spatial and temporal relations to make reasonable predictions on traffic conditions, helping vehicles choose the most appropriate route, especially in times of congestion.

For the preliminary contest, Baidu will provide the teams with a set of actual data of traffic conditions in a certain city from the past 50 weekdays for training. Each team will conduct data training using Baidus deep learning computing architecture, PaddlePaddle, to predict traffic every five minutes during the morning rush hour on the 51st weekday. Baidu will then judge each team on the accuracy of their traffic predictions.

This years ASC Student Supercomputer Challenge, the largest supercomputer contest in the world, is jointly organized by the Asia Supercomputer Community, Inspur, the National Supercomputing Center in Wuxi and Zhengzhou University. There are a total of 230 university teams from 15 countries and regions participating in the 2017 contest with the finalists announced on March 13th and the final competition held April 24th-28th.

The contest aims to inspire innovation in supercomputer applications and cultivate young talent. The era of intelligent computing is here and it is being driven by AI. High performance computing is one of the main technologies supporting AI and is facing changes and new challenges. It is with this that ASC has incorporated AI into the competition. It is hoped that more young university students can get involved in this trendy application more quickly and cultivate their enthusiasm for innovation.

For more information on the ASC17 preliminary contest, please visit http://www.asc-events.org/ASC17/Preliminary.php

Read the original:

Baidu Joins ASC17 Supercomputer Competition with AI Challenge - HPCwire (blog)