With the popularization and development of mobile Internet, and data leakage, phishing-related news is also increasing. According to the US credit reporting agencies benefit CMX (Experian), and in China, online fraud losses caused by the proportion of GDP accounted for 0.63%, ranking second in the world. How can we ensure that people and companies do big data security? Victoria Optional Technology CEO Xieying Lian told Sohu interview with science and technology, introduces the industry's traditional practices in the field of security - supervised learning algorithm, and she developed a method - unsupervised learning algorithm.
Xieying Lian introduced, the field of security industry has a certain particularity. Because fraudulent party will continue to change form, and often in the daylight side of the anti-fraud, fraud in the dark side, which makes the two sides in a dynamic game environment.
For network fraud, the traditional practice of domestic enterprises is invited to a black and white lists and other data to provide data services company, or set up a "supervised machine learning engine" within the enterprise. However, he supervised learning technologies logic is that first acquisition mode *** fraud Party, became the sample and experiential knowledge of machine learning engine, or directly to a list, and then let the machine learning engine to block off new ***.
Xieying Lian said, there is a drawback to this logic, as in the exercise under the drug will continue to evolve as viruses, fraudulent party will also produce a variety of changes after specialization. In this way, we supervised learning algorithm will continue to be in response to the state of motion lag, no way to effectively deal with the development model of network ***.
Xieying Lian developed unsupervised learning algorithm, it is to go explore the unknown *** models to deal with fraud pattern has not yet occurred. Fraud unsupervised algorithm computation of large data analysis on customer desensitization, so that early prevention and monitoring. The first user of a social dimension Optional Technology companies unfamiliar street, other customers include Alibaba, public comments and other companies.