Intel Collaborates With Baidu for Developing Advanced AI Technology Continuing long-term association

Last week in Beijing at Baidu Create, Intel has announced that it is working with Baidu on all AI hardware and software platforms. As both of them are working together since long time, Intel has been the key factor in case of developing advanced AI technologies for Baidu. In 2017, Baidu and Intel announced that they are continuing their long-term association for research and development of Nervana Neural Network for Training (NNP-T), a hardware accelerator optimized for deep learning.  NNP-T 1000 includes processor cores based on Intel’s ice lake architecture to optimize neural network. 

NNP-T and Xeon Scalable Processor

Other than NNP-T, Baidu also uses Intel Xeon Scalable processors that help in running the Baidu Brain. It is an AI platform which acts like a brain in all external and internal applications. The processor is used in over 100 AI services which includes facial recognition, natural language processing, voice processing and recognition. The 2nd generation Intel Xeon Scalable processor of Intel and Baidu help in optimizing the performance for natural language processing, workloads incorporating speech synthesis, visual applications and other aspects of the same.  

With the help of Xeon scalable processor infrastructure, NNP-T is expected to increase the AI efforts at Baidu, which in result will increase the deep learning training jobs. Baidu is working on PaddlePaddle (parallel distribution deep learning) for Intel NNP-T which will optimize its deep leaning framework.  It is the first China based deep learning framework, which will be an alternative to TensorFlow, an open source ML framework from Google. 

Vector Neural Network Instruction (VNNI) 

According to Intel, PaddlePaddle has some new improvements deliver dramatic performance which will enhance all the important functions such as image classifications, language translation, speech recognition, object detections. It is also the first framework to incorporate vector neural Network instructions. There are many other features, which are being introduced in the 2nd generation Intel Xeon scalable processors.