Huawei MindSpore AI computing framework is similar to what Google’s TensorFlow is all about. But the latter has an advantage as it is an open-source platform. Following its footsteps, Huawei has also made its Mindspore framework open-source. The company announced it on the second day of Huawei Developer Conference Cloud 2020.
The Chinese tech giant Huawei first announced the MindSpore AI computing framework in August 2019. It follows the three core development goals and they are easy development, efficient execution, and adaptable to all scenarios respectively.
As privacy has become a concern in today’s world, MindSpore only deals with gradient and model information, which has already been processed. That means it does not process the data itself and hence private user data remains protected even in cross-scenario environments. Further, it also has a built-in model protection technology for added security.
The MindSpore AI framework can be used in all scenarios, across all devices, edges and cloud environments. It also provides on-demand cooperation between them and its “AI Algorithm As Code” design helps developers to build advanced AI applications seamlessly. Because in a typical neural network for NLP (Natural Language Processing), it has 20% fewer lines of core code than other frameworks in the market and thus increases the efficiency of developers by at least 50%.
Huawei MindSpore AI framework not only supports the company’s own lineup of AI processors such as Ascend 910 but also other CPUs and GPUs available in the market.
UP NEXT: Huawei announces Ascend 910 AI chip and Mindspore AI framework
(Source)
Comments