Cicada to the global champion!Huazhong Big Team will take the picture to calculate the world's first

Wuhan Evening News August 18 (Reporter Chen Xiaotong Correspondent Wang Xiaoxiao) On August 15th, the most influential international competition in the field of calculation — 2022 map calculation challenge Graph Challenge results announced, the computing of computer colleges of Huazhong University of Science and Technology Since the team won the championship last year, two more participating works this year reached the top and won the global championship.

The two participating works of the Computing Team of the Computing of the University of China University of Science and Technology reached the summit at the same time, winning and winning the global championship.

According to reports, Graph Challenge has successfully held six sessions so far, and is one of the most influential international events in the field of map computing. The Graph Challenge competition includes three tracks: static map matching, dynamic diagram, and sparse neural network reasoning, which has attracted many well -known companies and colleges from the world, including Nvidia, Cylini, Huawei, Lawrenceerfo Moore. Laboratory, Carnegelon University, Washington State University, etc.

In the case of monopoly by well -known American schools and enterprises over the years, in 2021, the map computing team of Huazhong University of Science and Technology achieved the "zero breakthrough" of the tournament championship for China, breaking the Nvidia Company and the National Laboratory of Lawrence Loomore. The monopoly of the well -known American scientific research units on the competition champion. This year, the map computing team of Huazhong University of Science and Technology made a good achievement, and the two participating works won and won the global champion.

It is reported that the two entries who won the championship were completed by Sun Yufei and doctoral students Wang Qinggang, and doctoral students, Xu Shaoxian and master student Wu Minkang, respectively.

Wang Qinggang introduced that last year's championship plan laid a good foundation for this year, but there were still lack. The team members began to prepare the race earlier this year, chose sparse neural network reasoning as the main track, and repeatedly discussed and optimized the plan last year. The two entries that won the award start from different angles and adopt different solutions to improve the performance of sparse neural network reasoning. Through the joint efforts of the team, compared with the same track championship last year, the two participating works won the global championship with a good result of 8.12 times and 6.37 times performance.

Instructor Zheng Long said that graph (Graph) is a flexible data structure composed of a vertex and edge, which can naturally model the relationship between things and affairs in the real world. Therefore, many application scenarios in real life need to Use diagrams, such as financial trading maps, national grid maps, virus transmission maps, traffic road maps, etc. In order to tap the useful information from the huge graph data, the calculation technology of the figure was born, and it was widely used in financial fraud testing, grid failure detection, virus dissemination tracking, travel road planning, military intelligence analysis and other fields.

At present, the calculation of the figure has become a new outlet for domestic and foreign science and technology giants. However, the sparseness and irregularity of the reality diagram make the efficient large -scale diagram data processing and analysis more difficult. To this end, IEEE, Massachusetts Institute of Technology, and Amazon jointly founded the Graph Challenge international competition in 2017 to call on global researchers to develop new solutions to continuously improve the processing efficiency of maps and sparse data from real life.

Zheng Long said that for the first time in the computing team of the map in 2021, the team won the championship, and by 2022, the team won and won the championship, which is of great significance. The team will continue to cultivate in the field of figure computing, and maintain the advantages of the Chinese team in this field with better results.

[Edit: Yu Lina]

- END -