Qualcomm explains its AI vision: Focus on terminal-side artificial intelligence
Complementing the cloud with ubiquitous terminals
Qualcomm's vision for artificial intelligence
In the world we envision, terminals, machines, cars and everything will become smarter, simplifying and enriching our daily lives.
Since 2007, Qualcomm has been exploring spiking neural approaches to machine learning for computer vision and motion control, and has since expanded its research from biomimetic approaches to artificial neural networks, primarily in the field of deep learning. Today, Qualcomm has announced the acquisition of Scyfer BV , a spin-off company of the University of Amsterdam that focuses on cutting-edge machine learning technology, to enrich its talent team. Scyfer has created advanced solutions for companies in many different industries around the world, including manufacturing, healthcare, and finance.
Matt Grob, executive vice president of technology at Qualcomm Technologies, said:
We started basic research ten years ago, and our existing products now support many AI use cases: from computer vision and natural language processing to malware detection on various endpoints such as smartphones and cars. At the same time, we are researching broader topics such as AI for wireless connectivity, power management, and photography.
Many companies focus on running AI-related workloads in the cloud, but Qualcomm is focused on
implementing AI
on
the device side
, such as smartphones, cars, and robots,
to ensure that processing can be completed with or without a network or Wi-Fi connection. The advantages of on-device AI include instant response, improved reliability, enhanced privacy protection, and efficient use of network bandwidth.
Qualcomm will continue to advance artificial intelligence research and bring advanced machine learning technologies to the forefront of the industry. Examples of specific work done in this regard include:
-
Advancing neural network technology for semi-supervised and unsupervised training, such as generative adversarial networks (GANs), distributed learning, and privacy protection;
-
Network optimization for end-to-end applications, including compression, inter-layer optimization, sparse optimization, and other techniques to better utilize memory and space/time complexity;
-
and a dedicated hardware architecture designed to accelerate machine learning operations, resulting in better performance and lower power consumption on embedded terminals.
The acquisition of Scyfer also brings in Dr. Max Welling, the company’s founder and a renowned professor at the University of Amsterdam, to help Qualcomm further develop its R&D efforts. Dr. Welling will continue to serve as a professor at the University of Amsterdam, and the Scyfer team will continue to be based in Amsterdam. In 2015, Qualcomm and the University of Amsterdam also established a joint research lab, QUVA, which focuses on developing advanced machine learning technologies for mobile and computer vision. Qualcomm will continue to advance its collaboration with the University of Amsterdam to reinforce the importance of educating the next generation in this critical area.
Click Read the original article to learn about the key technologies of 5G!
京公网安备 11010802033920号