Invited Speaker---Dr. Masato Oguchi, Professor
Department of Information Sciences, Ochanomizu University, Japan
Masato Oguchi received B.E. from Keio University, M.E. and PhD from University of Tokyo in 1990, 1992, and 1995 respectively. In 1995, he worked as Center of Excellence (COE) researcher in National Center for Science Information Systems (NACSIS) - currently known as National Institute of Informatics (NII) in Japan. From 1996 to 2000, he was a research fellow at the Institute of Industrial Science, University of Tokyo. He stayed at Aachen University of Technology in Germany as a visiting researcher in 1998 – 2000. In 2001, he became an associate professor at the Research and Development Initiative in Chuo University. He has joined Ochanomizu University in 2003 as an associate professor. Since 2006, he has been a professor at the Department of Information Sciences, Ochanomizu University.
His major is network computing including Cloud Computing, IoT networks and security as well as data engineering like data mining and social data analysis based on machine learning. He has published many international conference papers in various fields, such as IEEE IPDPS, ACM ICS, and IEEE/ACM CCGrid in HPC field, IEEE ICC, IEEE WiMob, IEEE WCNC, and IEEE LCN in communications field, and IEEE ICDE, ACM HT, and IEEE Big Data Congress in data engineering field. He has served as organizing and program committee members in various international conferences such as IEEE AINA, IEEE ICPADS, ACM IMCOM, IEEE Data Com, IEEE ICC, IEEE Big Comp, and others.
Performance Evaluation of Trusted Data Mining Using Fully Homomorphic Encryption
To accumulate large amounts of data and calculate the statistics, data mining requires large capacity storage and high-performance computers in the computational environment. Therefore, uploading commercial data to cloud services is popular in general. An outsourcing data mining system, in which organizations can send their data to third-party cloud services, query them and obtain the statistics from the data, should be widespread in the future. However, since personal customer data owned by companies are confidential and proper security management is required, it is necessary to encrypt the data to be concealed from anyone, including the cloud service providers. For that purpose, data protection using fully homomorphic encryption (FHE) has been proposed for a client/server trusted data mining system using the Apriori algorithm. However, its execution requires much time due to the computational complexity of FHE calculations. Additionally, although frequent database updates occurred in the practical use of the system, the Apriori algorithm needs recalculation of the whole database at each update. To solve these problems, we propose the implementation of a distributed system using the FUP algorithm, which generates candidate item-sets efficiently while updating the database. We implement a master/worker distributed system to improve runtime to make the system more suitable for practical use, execute experiments and evaluate environmental system, which consists of general Linux servers for cloud computing platforms.