[San Francisco, USA, June 17, 2015] Huawei recently participated and won several awards in the Low-Power Image Recognition Challenge (LPIRC) at the Design Automation Conference 2015 (DAC 2015) held in San Francisco, working with two different partners to discover the best technology in both image recognition and energy conservation. Partnering with Tsinghua University, the joint team won the “Champion” and “Third” awards, and also the “highest accuracy with low energy” award. In the same challenge Huawei also partnered with the Institute of Automation, Chinese Academy of Sciences (CASIA) and won the “Second” and “least energy with good accuracy” awards. The Huawei joint teams both demonstrated strong technical solutions delivering high image recognition accuracy with low energy consumption. The aim of this challenge is to develop future cognitive computing systems.
Representing Huawei. Dr. Jun Yao, Principal Engineer, Huawei, said, “Cognitive computing has shown significant potential to impact future IT infrastructures. Huawei has recently started to research this field and explore multi-dimensional optimization technologies for both cognitive algorithms and platforms. Through leveraging the varied strengths of different research institutes, our research has been recognized at the LPIRC. This is our first time participating in an international competition for this field of technology, and we are honored to receive several awards in collaboration with Tsinghua University and CASIA. We will continue to work with industry and academic partners to deliver more innovations for cognitive systems.”
Huawei collaborated with their respective partners to leverage their strengths in different technical areas, including system prototype development, software and hardware co-design, full system tuning, and machine learning algorithms.
Professor Yu Wang, Tsinghua University, said “The LPIRC promotes rethinking in cognitive computing benchmarks, including the importance of energy conservation, accuracy of image recognition and implementing a balance between cognitive systems and Deep Learning algorithms.”
In recent years, technology has been developed enabling computers to recognize images through holistic algorithms, and cognitive computing. Deep Learning is a key technology to enable this and requires large datacenters to provide power for the many tasks involved to accurately identify objects in photos. The limit on data center power consumption is impeding the future growth of cognitive computing which has led to a focus on energy conservation. The global LPIRC challenge aims to discover the best technology for accurate image recognition and this year includes energy conservation as a key criteria for evaluating the winner.
Yung-Hsiang Lu, Co-Organizer, LPIRC, said, “The solutions tested by both Huawei and Tsinghua and Huawei and CASIA teams deliver outstanding quality and great results. We look forward to Huawei participating in future LPIRC events.”
Charles Alpert, Chairperson of the DAC executive committee, and Sani Nasif, President of Institute of Electrical and Electronics Engineers Council on Electronic Design Automation (CEDA), both commented that the Huawei teams achieved remarkable breakthroughs in the LPIRC.
Huawei will continue to invest in cognitive computing and collaborate with both international and domestic research institutes to establish a global collaborative ecosystem, and develop innovative cognitive system solutions.