- Conventional deep learning (machine learning) methods use one-frame images for convolutional bonding to extract features by learning data locations and points. Time-series Deep Learning convolutes pixels in multiple sequential frames.
- Whereas existing Convolutional Neural Network (CNN) technology has been shown to recognize motions with less than 60% accuracy, Time-series Deep Learning achieved more than 80% accuracy in recent tests.
Time-series Deep Learning will be exhibited at the NTT Communications Forum 2015, which will be held at the Prince Park Tower Tokyo in central Tokyo on October 8 and 9.BackgroundRecognition of objects and people in static images has dramatically improved with the use of deep learning technology, a form of artificial intelligence (AI). However, high-precision recognition of data which changes according to time series is still difficult. Meanwhile, the increasing use of monitors and other network cameras, as well as IoT devices such as cameras and sensors connected to the Internet, are expanding the availability of collectable image data. Businesses that can use this data are expected to grow as the preciseness of video analysis improves. About NTT Communications CorporationNTT Communications provides consultancy, architecture, security and cloud services to optimize the information and communications technology (ICT) environments of enterprises. These offerings are backed by the company's worldwide infrastructure, including the leading global tier-1 IP network, the Arcstar Universal One™ VPN network reaching 196 countries/regions, and 130 secure data centers worldwide. www.ntt.com | Twitter@NTT Com | Facebook@NTT Com | LinkedIn@NTT Com