Abstracts

  • Soohyun Ahn (Ajou) Corrected Nonparamtric Intervention Effect Estimator using Ranked Set Sampling with Binary Data

To estimate intervention effect or treatment effect is one of the main goals in many fields such as environmental, educational, social and biological sciences. In general, a study design relies on simple random sampling (SRS) to select a samples. To improve the efficiency of statistical inference about treatment effects, ranked set sampling (RSS) is used as a good substitue for SRS due to its cost effectiveness. In this study, we explore the corrected RSS estimator of the treatment effect with binary outcomes, where the treatment effect is defined as log odds ratio. A common issue in the practice is that one or more groups may have zero or small count leading to undefined log odds ratio and substantial bias in estimation. Therefore, we derive a theoretical correction formula that gives asymptotically unbiased estimator of the treatment effect. Further, we numerically compare the proposed correction method with a standard approach to bias reduction in traditional designs using SRS.

  • Youngwoo Choi (Ajou) ACMI, aims and progresses

In this brief talk, I would like to explain how ACMI was formed to fulfill various departmental needs, and how the center is changing both the students and faculty members of the department. Some of the experiences and progresses made so far will be also mentioned.

  • Yasuhide Fukumoto (Kyushu) Modeling compressible combustion flame and shear flow of a river by interfaces of velocity discontinuity

An interface across which velocity and/or thermodynamic quantities discontinuously vary serve as a universal model to mimic a wide range of phenomena of fluid flows, which is efficiently tractable by mathematical and numerical means. We address two stability problems of interfaces of discontinuity in tangential and normal velocities.

1) Effect of compressibility on stability of a planar front of premixed flame

2) Drag induced instability of surface of velocity discontinuity of a shallow-water flow

We discuss theirs applications to lean gas combustion and to natural disaster resilience.

  • Jae-Hun Jung (Ajou) Topological data analysis of complex flows: Applications to music and vascular flows

Characterizing complex flow behaviors is difficult in general but essential to understand the given system. In this talk, we briefly explain how topological data analysis (TDA) can be used to characterize complex flows and introduce a simple TDA methodology that we developed for the problem. We illustrate two applications, vascular flows for the diagnosis of vascular disease and Korean old music transcribed in Jeongganbo.

  • Naoyuki Kamiyama (Kyushu) Discrete optimization approaches to real-world problems

In this talk, I talk about the collaboration with Fujitsu Laboratories Ltd. on real-world problems, e.g., congestion, security, and resource allocation.

Especially, I focus on discrete optimization approached to these problems.

  • Soon-Sun Kwon (Ajou) Statistical Models for Longitudinal Data Analysis in Medical Research

In medical research, there exist a variety of data structures such as serial data sets with different follow-up intervals and points, longitudinal data sets with missing values, data sets with multiple measurements for individual subjects (for example, measurements from left and right sides), and so on. Specially, longitudinal data are used in statistical studies that accept many repeated measurements as well as the different time spans of the measurements between or within subjects. Furthermore, correct inferences can particularly be obtained by considering the correlation between repeated measurements within subjects. In this talk, I introduce statistical prediction models for handling longitudinal data in orthopedic data.

  • Tomoyuki Shirai (Kyushu) Dynamic determinantal point processes

Determinantal point processes (DPPs) have been receiving increasing attention in machine learning as a generative model of subsets consisting of relevant and diverse items. Recently, there has been a significant progress in developing efficient algorithms for learning the kernel matrix that characterizes a DPP. We briefly explain what DPPs are and how to learn DPP, and discuss a dynamical version of DPP and its application.

This talk is based on a joint work with T.Osogami (IBM Tokyo), R.Raymond (IBM Tokyo), A.Goel(Kyushu Univ.) and T.Maehara(RIKEN AIP).

Homomorphic encryption enables various calculations while preserving the data confidentiality. This gives a powerful tool for clients to securely outsource their data to the cloud, and the cloud can compute clients data on encrypted data (without decryption). In this talk, I introduce a typical homomorphic encryption scheme constructed from lattices. Moreover, I give some demonstrations for applications, such as secure biometric authentication and secure pattern matching.