AI US patent layout mainland even behind Taiwan

AlphaGo is amazing in the field of machine learning. In contrast, Taiwan still has a lot of room for progress in the patent layout in this field.

In the spring of 2017, the mysterious chess player Master went to the Chinese, Japanese and Korean Go game masters on the mainland Go website. The result of the match was 60 consecutive games. The Master confirmed that the Master is AlphaGo. This is followed by AlphaGo in March 2016. After defeating the 1st defeat of the South Korean Go 9 player Li Shizhen, in less than a year, AlphaGo once again showed significant progress in machine learning artificial intelligence calculation technology.

Research shows that machine learning has the ability to automatically learn and predict, and can help people solve problems in life and work, such as consumer behavior analysis, customer relationship management, stock investment transactions, credit fraud detection, malignant tumor detection, etc., Taiwan academic circles Research institutions and industry have invested in machine learning for many years. Compared with other countries or international companies, what are the opportunities and challenges for machine learning in Taiwan? We analyze machine-learned US-approved patents to gain an understanding of the differences in machine learning international R&D capabilities, and to gain an in-depth understanding of the machine-learning patent layout and its impact on the technology community.

Figure / Associated Press

First, the machine learns the distribution of countries in the US patent, and finds that it is mainly concentrated in the United States, followed by Japan and Germany. Taiwan’s share of only 1.1%, while the mainland is closely following, accounting for 0.4%. Secondly, the machine learns the distribution of US patents, and finds that international rights holders focus on computing technology, followed by control, measurement, and information technology management methods. Biotechnology, food chemistry, pharmaceuticals, medical technology, machine tools, and special machinery are still low. . Taiwan's right holders have a relatively high degree of control and measurement, and computing technology and information technology management methods are relatively low. Medical technology has several patents.

Furthermore, machine learning learns the distribution of US patents, and finds that international rights holders are concentrated in data processing systems for data identification, control systems, supervision or prediction purposes, followed by speech recognition, diagnosis, image data processing systems, and medical treatment. Supplies, power supply or distribution systems, robotic arms, marine propulsion devices, aircraft, weapon targeting tools, attack or defense equipment on ships, jet propulsion, electric vehicle power equipment, electric vehicle power equipment, wind turbines/engines, etc. In addition, the Taiwanese right holders have a relatively high control system, and the data processing system for data identification, monitoring or prediction purposes is relatively low. The speech recognition, diagnosis, and jet propulsion devices have obtained patent rights.

Finally, the machine learns the top 20 of the US patent international rights holders' relative R&D capabilities, and the industry's vendor attributes account for 100%. Among them, important patent holders are Microsoft, IBM, Google, Facebook, Sony, Oracle, Yahoo, AT&T, Amazon, Qualcomm, Samsung, Intel, Numenta, Health Discovery, Honda and other well-known manufacturers. In contrast, the property rights of Taiwanese property owners in the industry accounted for 40%, and the total proportion of research institutions and academics reached 60%. Important patent holders include Chengda, Industrial Research Institute, Asset Planning Association, and TSMC. , Hua Ya Ke, An Ke Sheng Medical, National Experimental Research Institute, Yishou University, Central University, National Taiwan University, Chunghwa Telecom, Hon Hai, Morning Star, Pingke University, etc.

It is worth noting that mainland index patent holders have successively obtained machine learning US patents. Among them, important patent holders include Qingda, Peking University, Northeastern University, Tencent, Huawei, Xiaomi, SMIC, Yinjiang, and Beijing Tsinghua University. Technology, etc., Alibaba, Baidu has not found a machine to learn US patents.

With the advent of the era of big data, the advantage of machine learning artificial intelligence calculus technology is that it can find out the rules from the huge amount of data to make predictions, and then make choices in a more efficient way, such as production management, inventory management, advertising marketing. , investment transactions, legal services, e-shopping, health care, logistics management, customer service, etc., the data show that Taiwanese rights holders have now obtained 43 US patents in the field of machine learning, but compared with the patented countries, such as the United States 2,350 patents, 356 patents in Japan, and 126 patents in Germany show that the research and development investment in the core technology of machine learning in Taiwan is relatively scattered, resulting in a relatively lack of machine learning patent layout.

Looking into the future, it is recommended that Taiwanese companies interested in investing in machine learning should consider combining the research institute's patented patent energy and technology foundation to effectively narrow the gap with leading players, and brainstorming to find innovative application gaps and further develop corresponding innovations. Technology, layout of emerging applications and services with potential for development, such as home robots, smart customer service, smart butler, virtual reality, biotechnology, energy storage technology, drones, electric vehicles, etc. (The author of this article is MIC Senior Industry Analyst)