Rule breach gets Baidu team sidelined in AI contest

A research team from Chinese search engine giant Baidu has been told to sit out an international artificial intelligence competi
A research team from Chinese search engine giant Baidu has been told to sit out an international artificial intelligence competition for sidestepping the rules

A research team from Chinese search engine giant Baidu has been told to sit out an international artificial intelligence competition for sidestepping the rules.

Baidu had boasted of besting rivals such as Google and Microsoft in an ImageNet Large Scale Visual Recognition Challenge (ILSVCR) contest in May to determine whose computing technology is better at recognizing and categorizing pictures.

Contest organizers this week posted a message saying that a Baidu team broke the rules in a way that gave it a potentially significant advantage.

Research teams were limited to making no more than two test submissions weekly to a database of images created for the challenge.

From November 28 of last year to May 13 of this year, there were at least 30 accounts used by a team from Baidu to submit to the test server at least 200 times "far exceeding the specified limit of two submissions per week," contest organizers said in an online message to members of the AI research community.

"This includes short periods of very high usage, for example with more than 40 submissions over 5 days from March 15th, 2015 to March 19th, 2015," the message read.

Feedback from the extra submissions may have allowed the Baidu team to better hone results and iprocesses, reasoned organizers of the contest based at a Stanford University lab.

The Baidu team was asked to take a year-long time-out from the contest, and told that it wouldn't be fair to compare its results with those of other competitors.

"Recently the ILSVRC organizers contacted the Heterogeneous Computing team to inform us that we exceeded the allowable number of weekly submissions to the ImageNet servers," Ren Wu of the team said in a note included with the notice from organizers.

"We apologize for this mistake and are continuing to review the results."

Artificial intelligence, or machine learning, has been an area of intense focus for Google, Microsoft, Facebook and others interested in having computers better understand how people think and what they are 'seeing' online.

Corruption probe

Baidu—often described as the country's equivalent of Google—said last month that it has launched anti-corruption investigations into its own employees after reports three department heads were being probed.

Baidu did not reveal any details of its inquiries, nor did it confirm the number or names of those being investigated.

"Internal anti-corruption action creates a just and transparent working environment, and a sound environment for workers' development," the company said in a statement to AFP.

"We call on all employees to respect company ethics and national laws, to create a healthy and just working environment."

Baidu fired five executives last November for bribery and illicit appropriation and dismissed four of its employees for bribery in August 2012, according to media reports.

© 2015 AFP

Citation: Rule breach gets Baidu team sidelined in AI contest (2015, June 4) retrieved 18 March 2024 from https://phys.org/news/2015-06-breach-baidu-team-sidelined-ai.html
This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no part may be reproduced without the written permission. The content is provided for information purposes only.

Explore further

China search engine Baidu reports slip in Q1 profit

214 shares

Feedback to editors