FB pixel

Intel and Baidu collaboration on neural network platforms could boost biometric performance

Categories Biometric R&D  |  Biometrics News

Intel and Baidu have formed a partnership to collaborate on hardware and software platforms for artificial intelligence, which could improve biometric facial and voice recognition performance, Forbes reports.

The companies are working together on research and development of a hardware accelerator optimized for deep learning, the Nervana Neural Network Processor for Training (NNP-T). The NNP-T 1000 was launched by Intel in 2017, with processor cores base on Intel Ice Lake architecture. The partnership, which extends a long history of cooperation, was announced at the Baidu Create AI developer conference.

Baidu is using Intel Xeon Scalable Processors to run the infrastructure for its Baidu Brain AI platform, which powers natural language processing, facial recognition, voice processing and recognition, and approximately 100 other AI services. The companies have also worked on optimizing the hardware to accelerate performance for speech synthesis, natural language processing, visual applications, and other workloads. Baidu is also optimizing its PaddlePaddle (Parallel Distributed Deep Learning) deep learning framework, which it offers as a alternative to Google’s TensorFlow, for Intel NNP-T.

Intel says PaddlePaddle is the first framework to integrate Vector Neural Network Instructions (VNNI) with the Intel Deep Learning Boost (DL Boost) acceleration features introduced with 2nd-generation Intel Xeon Scalable processors. The integration delivers significant performance improvement in image classification, speech recognition, language translation, object detection, and other applications, the companies say.

The companies are developing a secure data processing framework, MesaTEE, building on Intel Software Guard Extensions (SGX), and will also invest in technologies to protect AI algorithms used in the cloud and at the network edge.

Researchers from various organizations from MIT to Google have worked on developing methods to apply neural networks to improve biometric performance and efficiency in the cloud or for edge devices.

Related Posts

Article Topics

 |   |   |   |   | 

Latest Biometrics News


Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Most Read From This Week

Featured Company

Biometrics Insight, Opinion

Biometrics White Papers

Biometrics Events

Explaining Biometrics