Behind the Robot "goddess": Large data and artificial intelligence relationship geometry

Intermediary transaction SEO diagnosis Taobao guest Cloud host technology Hall

May 5 in Beijing, the Global Mobile Internet Conference (gmic 2014) scene, a known as the "sexiest" Robot "goddess" to earn the eye, that is, Osaka University of Japan, director of the Intelligent Robot Institute Ishiguro in the field of intelligent robot research, the latest results--actroid-f. "Goddess" In addition to lifelike appearance, "she" can also complete the nod, blink and other movements, and can carry out simple conversation and reaction, to achieve communication with mankind.

In the "goddess" of the beautiful appearance of the robot and the powerful features of the emotion, you will think of the implementation of its logic and the process is what? The robot "goddess" as a dazzling display of artificial intelligence technology, now hot big data technology is how to help it?

From the concept of artificial intelligence to the development of the present, the existing common artificial intelligence methods include artificial neural network, machine learning, knowledge performance, intelligent search, fuzzy logic and so on. However, it is not easy to make these methods have excellent performance, the premise is to have enough data samples and strong computational power. This in the early days of artificial intelligence, it is difficult to achieve, therefore, the development of artificial intelligence for many years to stay in the micro-innovation phase, and difficult to have a big breakthrough.

Now, the arrival of the big data age may give new impetus to the rapid development of artificial intelligence.

Taking machine learning as an example, the purpose of machine learning is to analyze and obtain laws automatically from the data, and to predict the unknown data by law. Currently widely used such as Internet search, spam filtering, machine translation, online advertising, handwriting recognition is a simple case of machine learning, but the intelligent machine learning is difficult to achieve relatively large. In particular, the complexity of the scene of communication with people, requires more preset conditions, machine recognition of the process of ambiguity in the problem will be flawed.

For data processing takes up more time and resources, this bottleneck restricts the development of artificial intelligence.

At present, domestic and foreign various large-scale websites have accumulated a large number of users of commodity transactions, search, social, personal feelings and other data. Large data technology has been able to efficiently and efficiently access, retrieve, classify, and statistic the data that are large, diverse, extremely low in value density and fast changing. It can even effectively analyze the large data collected and possessed, and further excavate the economic and social value of large data. But the intelligent one big problem is to the complicated scene reaction precision, the artificial intelligence if does not rely on the big data, will be difficult to have as.

Interestingly, the scientific and technological community also advocates "reciprocity", which is reflected in the interaction of artificial intelligence and large data. Some people in the industry said that artificial intelligence for many years of research results can further promote the development of large data. such as natural language semantic analysis, information extraction, knowledge representation, automated reasoning, machine learning, etc., these technologies are now being applied to the frontier of large data technology, mining the laws and values of large data, so as to provide support for human decision-making.

In addition to the common shopping, video recommendations and social maps, the recent hot googleglass and drone technology are also successful applications of AI technology in large data fields.

Last year, the uproar of the "Prism plan" is actually a large data application of large-scale experiments, its working principle is that the artificial intelligence system through the "key words + occurrence law" constantly to produce a large number of electronic traces, traces, traces of the search and analysis. The "key words" here are not just text that people type on the Web, but also sounds or images (now the technology that converts speech to text is not difficult, as we already know about Siri and so on). Ordinary voice can not only be restored to the text through the computer program, even some with individual unique color of the word rules and audio features can be recorded. The artificial intelligence program of "Prism plan" analyzes the electronic traces, tracks and traces of a large number of entities and virtual worlds, finds out the suspicious points which are inconsistent with ordinary people's daily life rules, and further tracks the suspicious points, so as to find out the problem or exclude suspicious suspects.

Nowadays, artificial intelligence has been creating commercial value and social value for human beings, but its potential can be further excavated with the help of large data, and not only that, the development of new technologies such as IoT and mobile Internet will bring more imagination space for artificial intelligence.