| ZHAO CHENGYAN | D, 中間発表 | 知能システム制御 | 杉本 謙二, 笠原 正治, 浦岡 行治(物質）, 松原 崇充, 小蔵 正輝 |
|
title: Stability Optimization and Stabilization for Positive Linear System via Convex Optimization
abstract: Positive linear system has received much attention these years due to the development of control theory and application. There are some related works on stability analysis and stabilization. However, when it comes to practice problems, people always suffer such kind of system in which their state matrix is not constant but time-varying. Even though, some researcher has proposed that by fixing the off-diagonal entries of the sate matrix, the optimal control of the time-varying positive linear system is feasible. However, this restriction prohibits people from modeling practical problems. In our research, we proposed a convex optimization framework for extending the result to all entries of the state matrix. Also, we test the effectiveness of our framework by applying to the dynamic resource allocation for product management process. Next, we continue our research on a particular case of time-varying positive liner system, which has great significance in modeling practical problem, called stochastic switched positive linear systems. Such kind of system contains a set of a positive linear system which is governed by a stochastic process. Since it is a stochastic system, the result for stability analysis and control is not suitable for research. Then, We introduce the mean stability result for a stochastic jump system. By utilizing the property of nonnegative matrix and optimization theory, we propose a framework for optimizing its mean stability. In the end, we show the feasibility of our result by solving the population control problem in the virus community. language of the presentation: English | |||
| 森 純平 | M, 2回目発表 | 知能システム制御 | 杉本 謙二, 笠原 正治, 松原 崇充, 小蔵 正輝 |
|
title: Interpolation of Markov process using Particle Swarm Optimization and Deep Learning
abstract: In this study, we present an optimization-based framework for interpolating Markov chains. We specifically formulate the interpolation problem as a nonlinear optimization problem, for which we use Particle Swarm Optimizations and Deep Learning. We illustrate the effectiveness of the proposed method by numerical simulations. language of the presentation: Japanese | |||
| ZAVIALOV IGOR | M, 2回目発表 | 数理情報学 | 池田 和司, 笠原 正治, 吉本 潤一郎 |
|
title: New approaches for Value-at-Risk estimation
abstract: Financial stability is a key factor for healthy economy. Value-at-Risk models (VaR) are powerful tools for financial risk management and are widely used by regulating authorities. The performance of risk assessment is difficult to analyze due to complex market conditions; therefore, new models have been continuously developing. In this presentation we present Markov Chain Monte Carlo (MCMC) method for VaR estimation and analyze its performance on real market data. We will also take a look into possible approaches for improving risk estimation and further developing of our method. language of the presentation: English | |||
| 髙田 大樹 | M, 2回目発表 | ソフトウェア工学 | 松本 健一, 笠原 正治, 石尾 隆, 畑 秀明 |
|
title: Constructing Large-scale Source Code Indices Using ZDD
abstract: Source code search is an indispensable technique for improving programmer productivity. In recent years, exploiting syntax information of source code in code search has attracted much attention. By extracting syntax information from the source code and applying it to search technology, it is possible to capture semantic features of the source code that text information could miss. Since this syntax information has a graph structure, there is a problem that calculation resources such as processing time and memory usage become larger than a text-based method of extracting information. In this research, graph information is efficiently expressed and indexed using ZDD to reduce the computational resources required for the search using syntax information. Furthermore, we aim to propose a realistic framework for handling syntactic information. language of the presentation: Japanese | |||