Our laboratory not only research general-purpose basic technology of machine learning, but also conducts applied research on machine learning in various fields closely related to NTT business, as well as in scientific fields.
Some applied research topics in our laboratory are as follows:
Civil activities in a city are not random because of regular and routine behaviors of people. Recently, sensor networks and mobile phone networks enable us to collect those activities as spatio-temporal big data. However, because such data exhibit complex phenomena and are often incomplete due to sensor corruptions or network troubles, it is hard to understand civil activities by just visualizing spatio-temporal flow in a city from obtained data. We propose a machine learning method that discovers latent patterns of spatial activity patterns and its corresponding temporal dynamics. Furthermore, our method can recover missing entries in data with extracted latent patterns. We demonstrate that our method can analyze automobile traffic data in a city and extract more interpretable patterns than existing pattern extraction methods.
Open House 2017 exhibition 02
We have developed a technology to dynamically compute and recommend comfortable migration schedules for customers in such as amusement parks based on spatio-temporal prediction of near-future congestion levels and resource demands by using real-time observation of the people flow and preferred attraction information. This technology is aimed at equalizing waiting time at attraction queues in a venue and maximizing customer satisfaction by real-time processing of spatio-temporal prediction of people flow and mathematical optimization of visitor’s migration schedules. It is also expected to support stable control of infrastructure and optimal resource management in and around venues such as leisure spots, airports, and commercial facilities.
Open House 2016 exhibition 04