Overview NTT Research, Inc. was established in Palo Alto, California—in the heart of Silicon Valley, USA—in April 2019. It comprises three laboratories, 1) Quantum science and computing, 2) Medical and health informatics, and 3) Cryptography and information security. The goal is to build a new human-resource ecosystem in the areas of advanced basic research that can fundamentally change the way we live and the way we work.
At the opening event held in July 2019, substantial encouragement and support were evident among the many notable guests from academia and the business worlds. Kazuhiro Gomi, President and Chief Executive Officer of NTT Research, Inc., a leader who thoroughly understands the importance of human networks, was asked what the prospects are for NTT Research, Inc.
Feature Articles: Communication Science for Achieving Coexistence and Co-creation between Humans and AI
Abstract Artificial intelligence (AI) has been making remarkable progress in recent years and has even been approaching the level of human performance for certain functions, but it still has its limitations. In contrast, human beings are highly advanced and complex, which is why they are also imperfect and prone to mistakes as reflected by their vulnerability to bias and illusions. This article introduces NTT initiatives in communication science to bring AI technology closer to a human level and to develop an even deeper understanding of human beings with the aim of closing the gap between AI and humans and achieving AI that can help people.
Abstract At NTT Communication Science Laboratories, we are researching information processing across different types of media information such as images, sounds, and text. This is known as crossmodal information processing. The point of crossmodal information processing is to create a common space, which is a place where multiple types of media data are associated. The common space enables us to realize new functions that have never existed before: new transformations between different media—such as creating images and descriptions from sound—and the acquisition of concepts contained in media information.
Abstract The human visual system differs considerably from person to person, and its ability varies with the context, task, and circumstances. To grasp the variability of visual ability in daily circumstances, we created two test batteries to easily measure visual abilities.
One is a simple visual test called a Tablet Test, which can be performed using conventional measurement methods. The other is a visual test battery called Shikaku no Mori that involves short video games. The gamification improves the enjoyability of the test compared with the conventional experimental method. The measurement time is approximately three minutes, and the accuracy of the test is comparable to that obtained in laboratory experiments, in which it often takes several hours to acquire data. The proposed test batteries would be useful for research in vision sciences as a method to investigate the diversity of visual ability and early detection of eye disease.
Abstract Body action such as walking is known to extend the subjective boundaries of peripersonal space (PPS; the space immediately surrounding our body) and to facilitate the processing of audio-tactile multisensory stimuli presented within the PPS. However, it is unclear whether the boundaries change when a sensation of walking is induced with no physical body motion. In this study, we presented several vibration patterns on the soles of the feet of seated participants to evoke a sensation of walking, together with an approaching sound toward the body. We measured reaction times for detecting a vibrotactile stimulus on the chest, which was taken as a behavioral proxy for the PPS boundary. Results revealed that a cyclic vibration consisting of lowpass-filtered walking sounds presented at the soles that clearly evoked a sensation of walking reduced the reaction times, indicating that the PPS boundary was expanded forward by inducing a sensation of walking.
Abstract Many difficulties arise in developing a dialogue system that can perform conversation in the manner of humans, even for casual conversations. Recent research on chat conversation has led to the development of dialogue systems that can respond to users in a wide range of topics, which was the first major challenge of chat conversation. However, it is still difficult to construct dialogue systems that can properly respond to user utterances according to the dialogue context, and this has often made users feel that the system did not understand what they said. In this article, we introduce our work on a chat dialogue system that has the ability to understand the dialogue context.
Abstract This article introduces CoCoNuTS, a technology for implementing an error correcting code (channel code) that achieves the efficient transmission limit known as the Shannon limit. It was once believed that a huge time complexity was necessary to achieve the Shannon limit for a given channel. However, practical channel codes that achieve the Shannon limit have recently been developed, but these codes achieve the Shannon limit only for a restricted class of channels. We have proven mathematically that we can construct codes achieving the Shannon limit with our CoCoNuTS technology. Furthermore, we have confirmed experimentally that the implemented codes outperform conventional codes for a channel where it is impossible to achieve the Shannon limit using the conventional codes.
Abstract Noninvasive glucose measurement without needle pricking is anticipated as a novel medical and healthcare application. We introduce here our research on the use of near-infrared photoacoustic spectroscopy and microwave dielectric spectroscopy for noninvasive glucose measurement using electromagnetic waves. We also present our recent work involving in vivo measurement. Both measurement techniques are based on optical and wireless components and system integration, which have been investigated in telecommunication system development at NTT.
Abstract The Open Networking Foundation (ONF) was founded in 2011, and since then, numerous open source communities focusing on software-defined networking (SDN) and network functions virtualization (NFV) have been launched and are becoming more active. This article first gives an overview of the open source communities concerned with SDN/NFV technologies and then describes the latest activities of the ONF and the initiatives undertaken by the NTT Group. The activities of the MEF (Metro Ethernet Forum) are also explained as an example of work pursued by the NTT Group.
Abstract NTT Communication Science Laboratories Open House 2019 was held at Keihanna Science City, Kyoto, on May 30 and 31, 2019. Around 1500 visitors enjoyed 5 talks and 30 exhibits, which included our latest research efforts in the fields of information and human sciences.
External Awards/Papers Published in Technical Journals and Conference Proceedings