Please note that JavaScript and style sheet are used in this website,
Due to unadaptability of the style sheet with the browser used in your computer, pages may not look as original.
Even in such a case, however, the contents can be used safely.
*** For immediate use June 22, 2012
Tokyo, June 22, 2012 - NEC Corporation (NEC; TSE: 6701) announced today the development of heterogeneous mixture learning technologies that automatically detect massive patterns hidden in big data. These technologies are expected to achieve high-precision predictions and anomaly detections that are difficult by handcrafted data analysis.
In recent years, the demand for analysis of big data collected from such tools as the Internet and specialized sensors is rapidly increasing. Today, there is a great deal of anticipation for technologies that help understand contemporary issues and even help predict future conditions. Currently, automatic extraction of patterns within big data is largely done by machine learning technologies (*1).
NEC's new heterogeneous mixture learning technologies learn multiple relationships hidden in big data and discover useful patterns. Among such patterns, they automatically select the appropriate one depending on the situation. This enables higher-precision prediction and anomaly detection in dynamically-changing environments than existing machine learning technologies, which often take just a single pattern into account.
For example, on the basis of extracted heterogeneous patterns, these technologies help property management companies to predict electricity demand even if relationships between electricity demand and factors such as surrounding temperature, day of the week or time of day are constantly changing. In the medical field, these technologies may also be useful for detecting abnormal patterns from "lifelog" data, potentially resulting in early detection of asymptomatic illnesses.
Going forward, NEC will continue to drive the development of technologies and systems that capitalize on the possibilities of big data.
NEC will present results from this development on June 28 at the 29th International Conference on Machine Learning (ICML 2012) in Edinburgh, Scotland from June 26 to July 1, 2012.
*1) Machine learning technologies:
Technologies that extract information from data, including useful patterns, rules, knowledge representations, evaluation criteria and others.
Takehiko Kato
NEC Corporation
+81-3-3798-6511
E-Mail:t-kato@cj.jp.nec.com
Joseph Jasper
NEC Corporation
+81-3-3798-6511
E-Mail:j-jasper@ax.jp.nec.com
Share: