Kakenhi are funds that provide broad support for scientic research based on the free ideas of the researchers themselves, and covers a wide range of academic studies spanning from basic to applied research. Both faculty members and researchers actively apply to Kakenhi for grants, and many are approved. The grants obtained from Kakenhi are also distributed to researchers in other institutions (co-investigators) for collaborative research work.
Similarly, many NII faculty members also participate as co-investigators in the Kakenhi-funded projects of researchers at other institutions.

Applications Accepted(FY2021)

No. of applications accepted Amount
(in thousands of yen)
Project Leader
(Principal Investigator)
69 397,950
(Other institutions → NII)
59 84,988

[Model Cases of Research Funded by Kakenhi]

Research on master biometric information protection and utilization platform

 Grant-in-Aid for Scientific Research (A) 

Principal Investigator: ECHIZEN, Isao
Professor, Information and Society Research Division

With the proliferation of high-performance cameras and microphones, biometric data defining human faces, voices, gaits, fingerprints, veins, irises, and other characteristics can now be captured and recorded remotely and shared in cyberspace. This poses the threat of spoofing," i.e., breaches of biometric authentication to commit fraud or identity theft. For this kind of spoofing, it was previously necessary to restore the biometric data of a person from the captured image or recorded audio, but now with advances in machine learning, it is possible to generate biometric data that can be recognized as matching multiple persons from publicly available biometric data sets (i.e., master biometrics) without restoring biometric data of a specific person.
This study aims to establish a biometric data protection and utilization platform that prevents spoofing by detecting master biometric data while at the same time continuing to guarantee the usefulness of biometric data sets used to generate such information and "neutralizing" the inherent threat posed by biometric data sets.


Explainable next-generation media forensics technologies based on fake media detection and automatic fact verification

 Grant-in-Aid for Scientific Research (A) 

Principal Investigator: YAMAGISHI, Junichi
Professor, Digital Content and Media Sciences Research Division

In the current age of "infodemics," fake media in the form of video, audio, and text that resemble the real thing can be generated easily with machine learning, resulting in floods of fake news and other inaccurate information.
To counter this threat, this study proposes a pioneering next-generation media analysis technology to help ensure the publication of accurate media and information and support effective decision-making. It firstly proposes a liveness detection method that improves the explanatory power of authenticity judgments by identifying and indicating the falsified areas and methods of fake media as evidence. Next, the study proposes a new detection method that, in principle, incorporates the ability to deal with unknown fake media generation methods, so that the method is capable of robustly dealing with constantly changing media generation techniques.
An approach to learning this detection method is also suggested. Additionally, the study aims to make advances in automatic fact verification, for automated fact checking, and to integrate this with media analysis technology.


Robust AI by integration of knowledge representation and machine learning

 Grant-in-Aid for Scientific Research (A) 

Principal Investigator: INOUE, Katsumi
Professor, Principles of Informatics Research Division

In Artificial Intelligence (AI) research, pattern recognition capabilities have improved dramatically in recent years, thanks to advances in the development of machine learning (ML). However, for advanced intelligence tasks involving symbolic processing, knowledge representation and reasoning (KR) have been used.
This study integrates the two technologies of ML and KR, which up to now have been studied independently, to establish a technological foundation for building a next-generation AI system that is both explainable and robust.
For this purpose, three research goals were set:
(1) to improve the explainability and updatability of ML methods by deploying KR techniques;
(2) to develop robust KR methods supported by ML techniques;
(3) to develop groundbreaking AI applications through the integration of ML and KR.

Researches on Model-aided Learning Approaches for Reliable Realtime Control in Future Wireless Systems

 Grant-in-Aid for Scientific Research (A) 

Principal Investigator: JI, Yusheng, Professor
Information Systems Architecture Science Research Division

To support advanced applications and intellectual innovation in the Super-smart Society, it is necessary to further enhance the functionality, performance and reliability of the information and communication service infrastructure. Using conventional model-based approaches, it becomes more and more difficult to solve the centralized and/or distributed control problems in multidimensional space of dynamically configured wireless systems.
In this research, we study signal processing, resource allocation, interference mitigation, autonomous access control, and mobility control problems in wireless communication systems, by means of integrated approaches based on mathematical models and machine learning.
By comprehensively considering spatio-temporal constraints on network resources and seamlessly coordinating communication, computation, storage, and control functions, we aim to achieve highly reliable real-time processing capability at an end-to-end basis.

Exploration of super multi-view construction techniques for creating light fields in a real space in which visual obstacles are cancelled out

 Grant-in-Aid for Challenging Research (exploratory) 

Principal Investigator: KODAMA, Kazuya
Associate Professor, Digital Content and Media Sciences Research Division

Although plagued by pillars and walls that greatly obstruct views, cramped multitenant buildings have been diverted as inexpensive community spaces, becoming sustainable centers of community that powerfully support new countercultural activities such as theater, music, and  lm--from longstanding live music venues to theaters where numerous idol groups have been nurtured.
Now, in the new era of social distancing required for pandemic control, it is essential to resolve these visual problems to enable more efficient use of compact urban spaces by further recycling cramped city spaces.
This study sets out to construct a super multi-view system for freely inputting and outputting light rays through the space in front of and behind shielding objects, for the purpose of achieving a virtual transparency of visual obstacles.

Prevention from Automated Analysis Services with Object-Level Adversarial Examples

 Grant-in-Aid for Early-Career Scientists 

Principal Investigator: LE, Trung-Nghia
Project Assistant Professor, Information and Society Research Division

Data analysis services are typically trained on images crawled from social networks without authorization. As a result, users need solutions to protect their privacy. Conventional adversarial examples are applied to entire digital images to protect contents, but it leads to unnatural results in human vision. Results also are fragile and easily disabled by transformations and compression during sharing and storing.
This research aims to explore object-level adversarial examples to protect the private information of users from data analysis services when they upload and share photos to social networks. We plan to against object localization, landmark recognition, and vision-language systems to prevent analyzing users' information without authorization.
We expect our solutions to automatically identify manipulable regions in images to minimize the effect on image quality, make protected images look inconspicuous, and appear natural to human vision.

Modeling of the motor recovery process and optimization of rehabilitation strategy using VR

 Grant-in-Aid for Scientific Research on Innovative Areas (Research in a Proposed Research Area) 

Principal Investigator: INAMURA, Tetsunari
Associate Professor, Principles of Informatics Research Division

Rehabilitation for motor dysfunction has involved a lot of subjective elements and estimations on the part of physical therapists, who formulate rehabilitation policies by predicting the recovery conditions of the patients' physical functions. This research aims to realize a system that provides an optimum rehabilitation program in response to the individual patient's conditions by optimizing the interaction process between the physical therapist and the patient.
Our goal is to model the rehabilitation process Y=f(X), where the rehabilitation strategy f is applied to the current motor function state X - the current motor disability - to change the current motor function state into the desired motor function state Y. We can expect physical therapists to decide on highly effective rehabilitation strategies with the proposed rehabilitation process model.

Study on Distributed Consensus by Using Synchronizing Vibration

 Grant-in-Aid for Scienti c Research (B) 

Principal Investigator: SATOH, Ichiro
Professor, Information and Society Research Division/a>

We will try to make distributed consensus more efficient by using a mechanism inspired from synchronization phenomena in vibrating systems in nature (e.g., the synchronization of the expansion and contraction cycle of the heart muscle and the transmission cycle of fireflies) into distributed systems. Distributed consensus serves as the basis for a variety of existing distributed algorithms, but it is known that the cost of reaching consensus increases significantly when multiple computers simultaneously demand consensus be reached, while simultaneously making other demands, because that many demands at one time cause the distributed consensus processing to be reworked. On the other hand, since most distributed consensus approaches tend to repeat multicast communication and replies to it in a sequential manner, thus resembling synchronization phenomena in vibrating systems in nature, we will propose and implement a method to introduce the synchronization mechanism in nature into distributed systems and will evaluate the proposed approaches.