AI development challenge | Microsoft proposed five major architecture

AI out of the experimental stage, began to enter the field of application, a few days ago Microsoft Asia Research Institute Hong Xiaowen pointed out that the implementation of the application of AI, Microsoft put forward five capacity-level architecture, including computing and memory, perception and visual auditory ability, cognitive insight and reasoning Planning to decision-making ability, creativity and ability to perceive the theory of relativity, and ultimately have the intelligence of others, and continue to study its applications to help companies to break through these difficulties.

Hong Xiaowen pointed out that AI is facing four major challenges in its development, including the sharp increase in demand for AI professionals in the industry but the serious imbalance between supply and demand; the lack of AI platform and tools for business owners to predict the future of AI and how to use AI; In order to support the simple and smooth application of enterprises and individuals; and the relevant AI data is not easy to obtain and difficult to extract valuable data therefrom.

Microsoft's five major architectures will help vendors speed up the introduction of AI. At the most basic level of computing and memory architecture, Microsoft aims to popularize AI and build an AI framework, AI computing, data platform, and robot on the Azure platform Framework, Cognitive Services, Advanced Machine Learning, Deep Learning and Image Recognition Capabilities in Tier 2 Perception and Vision Auditory Architecture, such as Microsoft's Top 5 AI Cognitive Services APIs for Enterprise Applications, including Visual, Speaking , Language, knowledge, search, such as Microsoft How-Dld.net can identify people's visual age, "Twins Or Not?" Face similarity test, Microsoft's photo recognition AI "CaptionBot", just upload any one Picture, you can automatically identify and explain the contents of this photo, including the person in the photo who is what kind of postures, scenic spots or state.

In the third layer of cognitive insight and inference-planning to decision-making architecture, Microsoft hopes to accelerate data-driven insights by building simple and powerful insights and experiencing experiences such as Microsoft AI Robot Little Ice, hosting programs, Writing poetry, singing, writing news and other capabilities, the fourth layer of the creative structure, from the production of poetry or lyrics and fragment melodies, to be able to distinguish and be able to create image style conversion such as Microsoft Pix camera APP service, you can Transforming a photo into a work of art masterpiece can also have the same functions and effects as the image style conversion.Style level style transfer can recognize that the picture is original or composite and can accurately convert the desired style to the photo Converted into a style very similar to the original artwork.

For the development of AI worries, Hong Xiaowen pointed out that the development of AI is at a turning point, the computer is also continuous improvement, through more data learning, in the near future can handle more complex and routine tasks; however, to Real problem solving requires Super Intelligence (SI), which is a combination of AI and HI Human Intelligence. It does not need to worry too much about the development of AI. Human beings can use AI in the past with technology. Assisted by AI to increase productivity and create higher business value.