NEC Central Research Laboratories is conducting research to discover hidden regularities in big data, to use in detecting anomalies and future prediction, as well as in resolving various issues in social and corporate systems. More concretely, by targeting large-scale complex systems beyond human capacity to control, we are aiming not just to understand the current situation and make future predictions, but also to conduct optimal action planning.
NEC core technologies: Heterogeneous mixture learning
Use of knowledge obtained from analysis of big data is now becoming essential for business. However, big data is difficult to analyze because often data that follow different patterns and regularities are mixed and stored together. How to analyze this heterogeneous mixture of data is a major point of focus.
In response, NEC has developed the world's first "heterogeneous mixture learning" technology in the form of a unique big data analysis engine that combines high accuracy with excellent interpretation capabilities. The engine works by automatically disaggregating and extracting multiple regularities that are mixed in the observed data.
Features of heterogeneous mixture learning
Heterogeneous mixture learning automatically discovers specific regularities from connections between various types of diverse data that have been mixed together, and can switch among reference rules in accordance with the data under analysis. This enables high accuracy predictions and anomaly detection even for data where the regularities change—data that is traditionally difficult to analyze using conventional machine learning that discover and use only a single regularity as its reference rule.
Examples of applications of heterogeneous mixture learning
- Heterogeneous mixture learning is being provided to many clients in the form of solutions such as the three below.
Predictive Analytics Solution for Fresh Food Demand
- To enable more efficient ordering, eliminate lost sales opportunities, and minimize disposal rates for perishable food products
Electric Power Demand Predictiong Solution
Predictive Analytics Solution for Demand of Repair Parts
November 11, 2015
NEC announces enhancement of its AI-related business
Data Mining TG supports the realization of Solutions for Society using Only 1, No.1, AI-related technologies (heterogeneous mixture learning, recognizing textual entailment, predictive robust optimization framework). Please see the press release for details.
- NEC strengthens AI-related business (Press release)
- NEC develops Artificial Intelligence technology that makes decisions and plans based on predictions using Big Data (Press release)
June 19, 2014
NEC has achieved fast and high-precision analysis of super-large-scale data by utilizing its Heterogeneous Mixture Learning Technologies that discover massive patterns hidden in big data to automate processes that data analysts with advanced expertise have until now performed manually.
NEC has developed new analysis techniques that automate trial-and-error processes that analysts have traditionally had to perform manually in order to discover patterns in analyzed data, namely "partitioning data" based on conditions such as day of the week or weather and "combining factors" that are important in making forecasts. This enables super-large-scale demand forecasting (for example, sales forecasts by store and product, energy demand forecasts, etc.) on millions of analysis targets, something which has been limited by conventional manual techniques.
Please see the press release for more details on this technology.