What is artificial intelligence?
Artificial intelligence (AI) is a new technical science, which aims at researching and developing the theory, technology and application system that simulates, expands and expands people. Artificial intelligence is divided into three stages: computational intelligence, perceptual intelligence and cognitive intelligence. The first is computational intelligence. Robots began to calculate and transmit information like humans, such as neural networks and genetic algorithms. Secondly, perceptual intelligence, including vision, voice and language, the machine begins to understand, make judgments and take some actions, such as a speaker that can understand voice; The third is cognitive intelligence. Machines can think like people and take action on their own initiative, such as driverless cars and autonomous robots that drive completely independently.
What is big data?
Bigdata refers to a collection of data that cannot be captured, managed and processed by conventional software tools within a certain time range. It is a massive, high-growth and diversified information asset, which needs a new processing mode to have stronger decision-making, insight and discovery, and process optimization ability. Big data takes data as the core resource, collects, stores, processes, analyzes, applies and displays the generated data, and finally realizes the value of the data.
Big data and artificial intelligence complement each other.
The accumulation of big data provides fuel for the development of artificial intelligence. IDC and Seagate Technology released the white paper "Data Age in 2025". The report shows that by 2025, the total global data will reach 163ZB. This means that the total data in 2025 will increase by more than 10 times compared with the total data generated globally in 20 16 years. Among them, the total amount of data for data analysis will increase 50 times compared with 20 16, reaching 5.2ZB (10 trillion bytes); The total amount of data belonging to cognitive system will reach 100 times. The explosive growth of data promotes the germination and growth of new technologies, and provides rich data soil for deep learning methods to train computer vision technology.
Big data mainly includes collection and preprocessing, storage and management, analysis and processing, visual computing and data security. It has the characteristics of expanding data scale, diverse types, fast generation speed, high processing capacity, strong timeliness, strict reliability, high value but low density, and provides rich data accumulation and training resources for artificial intelligence. Taking the number of training images used for face recognition as an example, Baidu's training face recognition system needs 200 million face portraits.
Data processing technology promotes the improvement of computing power. The field of artificial intelligence is rich in massive data, and the traditional data processing technology is difficult to meet the requirements of high intensity and high frequency processing. The emergence of AI chips has greatly improved the efficiency of large-scale processing of big data. At present, GPU, NPU, FPGA and various AI-PU special chips have appeared. Traditional dual-core CPU takes days or even weeks even for simple neural network training, while AI chip can improve the operation speed by about 70 times.
Algorithms make a lot of data valuable. Whether it is Tesla's driverless or Google's machine translation; Whether it's Microsoft's Xiao Bing or Intel's Precision Medicine, you can see the figure of "learning" a lot of unstructured data. The development of technologies such as "deep learning", "reinforcement learning" and "machine learning" all promote the progress of artificial intelligence. Taking computational vision as an example, as a field with complex data, the recognition accuracy of traditional shallow algorithms is not high. Since the emergence of deep learning, on the basis of finding suitable features, the machine recognition accuracy of almost all images representing computer vision has increased from 70%+ to 95%. It can be seen that the rapid evolution of artificial intelligence not only needs theoretical research, but also needs a lot of data as support.
Artificial intelligence promotes the deepening of big data applications. Driven by the exponential growth of computing power and high-value data, intelligence with artificial intelligence as the core is constantly expanding the breadth of technology application, expanding the depth of technology breakthrough and continuously improving the speed of technology landing (commercial realization). For example, in the new retail field, the combination of big data and artificial intelligence technology can improve the accuracy of face recognition, and merchants can better predict monthly sales; In the field of transportation, the combination of big data and artificial intelligence technology, intelligent traffic flow prediction based on a large number of traffic data, intelligent traffic guidance and other artificial intelligence applications can realize the intelligent control of the overall transportation network; In the field of health, the combination of big data and artificial intelligence technology can provide more convenient and intelligent medical services such as medical image analysis, auxiliary diagnosis and treatment, and medical robots. At the same time, at the technical level, big data technology has basically matured and promoted the progress of artificial intelligence technology at an alarming rate; At the industrial level, intelligent security, autonomous driving and medical imaging are all accelerating.
With the rapid application and popularization of artificial intelligence, big data continues to accumulate, and algorithms such as deep learning and reinforcement learning are constantly optimized. Big data technology will be more closely integrated with artificial intelligence technology, with the ability to understand, analyze, discover and make decisions on data, so as to obtain more accurate and in-depth knowledge from data, tap the value behind data, and give birth to new formats and new models.