Hidemi Yamachi

Multi-objective genetic algorithm for solving N-version program design problem

Abstract N -version programming (NVP) is a programming approach for constructing fault tolerant software systems. Generally, an optimization model utilized in NVP selects the optimal set of versions for each module to maximize the system reliability and to constrain the total cost to remain within a given budget. In such a model, while the number of versions included in the obtained solution is generally reduced, the budget restriction may be so rigid that it may fail to find the optimal solution. In order to ameliorate this problem, this paper proposes a novel bi-objective optimization model that maximizes the system reliability and minimizes the system total cost for designing N -version software systems. When solving multi-objective optimization problem, it is crucial to find Pareto solutions. It is, however, not easy to obtain them. In this paper, we propose a novel bi-objective optimization model that obtains many Pareto solutions efficiently. We formulate the optimal design problem of NVP as a bi-objective 0–1 nonlinear integer programming problem. In order to overcome this problem, we propose a Multi-objective genetic algorithm (MOGA), which is a powerful, though time-consuming, method to solve multi-objective optimization problems. When implementing genetic algorithm (GA), the use of an appropriate genetic representation scheme is one of the most important issues to obtain good performance. We employ random-key representation in our MOGA to find many Pareto solutions spaced as evenly as possible along the Pareto frontier. To pursue improve further performance, we introduce elitism, the Pareto-insertion and the Pareto-deletion operations based on distance between Pareto solutions in the selection process. The proposed MOGA obtains many Pareto solutions along the Pareto frontier evenly. The user of the MOGA can select the best compromise solution among the candidates by controlling the balance between the system reliability and the total cost.

Design of a Multi-Robot System Using Mobile Agents with Ant Colony Clustering

This paper presents a framework for controlling mobile multiple robots connected by communication networks. This framework provides novel methods to control coordinated systems using mobile agents. Instead of physical movement of multiple robots, mobile software agents can migrate from one robot to another so that they can minimize energy consumption in aggregation. In some applications, it is desirable that multiple robots draw themselves together automatically. In order to avoid excessive energy consumption, we employ mobile software agents to locate robots scattered in a field, and make them autonomously determine their moving behaviors by using a clustering algorithm based on the Ant Colony Optimization (ACO). ACO is the swarm intelligencebased method that exploits artificial stigmergy for the solution of combinatorial optimization problems. Preliminary experiments have provided a favorable result. In this paper, we focus on the implementation

Integrating Uncomfortable Intersection-Turns to Subjectively Optimal Route Selection Using Genetic Algorithm

Route selection is one of the most important problems for a car navigation system. Given a pair of origin and destination, there are many possible routes. Most current car navigation systems propose the shortest path from the origin to the destination. Selecting the shortest path is not a hard problem, but the shortest path is not always what the user wants; what the user really wants to have is the most comfortable route for him or her to drive. In other words, the driver wants to have a car navigation system to propose the subjectively optimal route for him or her. Finding such a route requires enumerating all the possible routes, and is known as a NP-hard problem. In order to reduce computational complexity, we have employed a GA to find a (subjectively) quasi optimal route for the driver. In this paper, we report our attempt to integrate uncomfortable-turns in to the conditions of our GA-based route selection algorithm. The numerical experiments demonstrate the feasibility of our GA-based route selection method.

Searching Pareto Solutions of Bi-objective NVP System Design Problem with Breadth First Search Method

The n-version programming (NVP) is a programming approach for constructing fault tolerant software systems. This approach employs functionally equivalent, yet independently developed software components. Each component is independently designed and implemented to meet the same system requirements. The same set of inputs is supplied to all n versions and they produce their own results. A decision mechanism then gathers the results from n versions and determines the result to be delivered to the user. In general, the NVP design problem has been formulated as the single-objective problem maximizing the reliability under the constraint of the cost limit. For such formulations, the dynamic programming or the genetic algorithms have been used. They are, however, time-consuming and do not guarantee to produce the least-dominated solutions. In this paper, reformulate NVP design problem as the multi-objective optimization problem that seek Pareto solutions, and we then propose an algorithm that employs the BFS (breadth-first search) method to find the Pareto solutions under practical computation time

Dijkstra beats genetic algorithm: Integrating uncomfortable intersection-turns to subjectively optimal route selection

Route selection is one of the most important problems for a car navigation system. Given a pair of origin and destination, there are many possible routes. Most current car navigation systems propose the shortest path from the origin to the destination. Selecting the shortest path is not a hard problem, but the shortest path is not always what the user wants; what the user really wants to have is the most comfortable route for him or her to drive. In other words, the driver wants to have a car navigation system to propose the subjectively optimal route for him or her. Finding such a route requires enumerating all the possible routes, and is known as a NP-complete problem. In order to reduce computational complexity, we employed a GA to find a (subjectively) quasi-optimal route for the driver. In the previous paper, we reported our attempt to integrate uncomfortable-turns to our GA-based route selection algorithm. Even though GA provided a quasi-optimal route in most cases and certainly much faster than exhaustive search, we have found that the classical Dijkstra algorithm can be used to find subjectively the most comfortable route by changing the data structure. In this paper, we show how the data structure that represents routes can be modified to assist Dijkstras algorithm to find subjectively optimal driving route. The numerical experiments demonstrate the feasibility of our route selection based on Dijkstra algorithm with extended data structure.

Aggregating multiple robots with serialization

This paper presents the design of an intelligent cart system to be used in a typical airport. The intelligent cart system consists of a set of mobile software agents to control the cart and provides a novel method for alignment. If the carts gather and align themselves automatically after being used, it is beneficial for human workers who have to collect them manually. To avoid excessive energy consumption through the collection of the carts, in the previous study, we have used ant colony optimization (ACO) and a clustering method based on the algorithm. In the current study, we have extended the ACO algorithm to use the vector values of the scattered carts in the field instead of mere location. We constructed a simulator that performs ant colony clustering using vector similarity. Waiting time and route to the destination of each cart are made based on the cluster created this way. These routes and waiting times are conveyed by the agent to each cart, while making them in rough lines. Because the carts are clustered by the similarity of vectors, we have observed that several groups have appeared to be aligned. The effectiveness of the system is demonstrated by constructing a simulator and evaluating the results.

Pareto Distance-based MOGA for Solving Bi-objective N-Version Program Design Problem

N-version Program (NVP) is a programming approach to fault tolerant software systems. It employs functionally equivalent, yet independently developed software components. We formulate the optimal design problem of NVP system to a biobjective optimization model, i.e., maximizing the system reliability and minimizing the system total cost. We use a Multi-Objective Genetic Algorithm (MOGA) to solve multi-objective optimization problems, however, it requires an appropriate mechanism to search Pareto solutions evenly along the Pareto frontier as many as possible. In our MOGA, we employ the random-key representation and the elitism and Pareto-insertion based on distance between Pareto solutions in the selection process. The proposed MOGA will obtain many Pareto solutions along the Pareto frontier evenly

A solution method employing a multi-objective genetic algorithm to search for pareto solutions of series-parallel system component allocation problem

We discuss the optimal system component allocation problem for series-parallel systems with interchangeable elements. A series-parallel system consists of subsystems that are connected in series and each subsystem consists of components in parallel. There are some heuristic methods to obtain quasi optimal solutions for the component allocation problem of series-parallel systems. Because this problem is one of the NP-complete problems, it is difficult to obtain the exact solutions for large scale problems. We had formulated this problem as a multi-objective optimization problem minimizing the system cost and maximizing the system reliability, and proposed an algorithm that obtains the exact solutions of the problems in an efficient way. The algorithm utilized the depth-first search method to eliminate useless searches and employs the branch-and-bound method to obtain the Pareto solutions. According to the results of our numerical experiments, the algorithm searches the Pareto solutions in practical time for not so large problems. In order to solve larger problems, in this paper, we propose a Multi-Objective Genetic Algorithm (MOGA). In comparison with the exact solution method we had proposed and the MOGA method, we assure the MOGA method produces the best compromised solutions and the later can solve large scale problems. We have conducted experiments for reasonably large scale problems and have demonstrated reasonably good performance of our method.

Developing Evaluation Criteria for a Service-Learning Course in Computer Science Education

Service learning, by which students apply what they have learned in a specific field to voluntary service work, continues to garner attention as an educational method for nurturing multiple diverse skills. To the best of our knowledge, a reliable evaluation method has not yet been established for service learning. In this study, we develop evaluation criteria for Information Volunteer, a service-learning course started in 1997 wherein students engage in information-based educational activities at nearby schools. In this study, we analyzed (1) the types of problems students faced in during volunteering, (2) how students solved (or failed to solve) these given problems, and (3) what students found to be meaningful in practice. We used participant observation and group interviews as research methods as well as a case-code matrix, a qualitative data analysis method. We generated 30 cases that we organized into the following three categories: problems, provision, and what they learned from the activity. Each of these categories was further divided into the following subcategories: communication, office work, and teaching. In our observations and interviews, we found that undergraduates were unable to solve all problems. Furthermore, students demonstrated different levels of performance, which we concluded might stem from disparities in the academic abilities of the students as well as the variety of activities from school to school. From our results, we developed assessment criteria in the form of a rubric for assessing student performance in the given subject areas.

Practices of Self-paced Learning with Prior Learning Materials and Exercises in First-year Programming Courses

工学教育(J.of JSEE), 65–5(2017) 1.はじめに 1.1 研究の背景 工学系大学の情報系学科における初年次プログラミン グの授業では, プログラミング学習の経験者と初学者が 混在しており,学生の学習履歴に合わせたクラス編成や 授業運営に工夫が必要となる.さらに,プログラミング 学習には,複数の学習項目を理解しながら,継続的な学 習が求められるため,学習意欲や理解度の差により,授 業の進度に遅れてしまう学生が多くいるのが実情となっ ている.このような多様な学生の状況に加え,専門基 礎科目に位置づけられるプログラミング科目は,後継の 専門科目につながる知識や技術を習得する役割も担って おり,教育の質の確保と向上に向けた授業方法の改善が 必要となっている. このようなプログラミング教育の問題に対して,学生 が主体的に授業に参加する能動的学習の仕組みを取 り入れた授業の実践報告が行われてきている.著者 らの所属する日本工業大学工学部情報工学科では,2013 年度より,初年次向けのプログラミング科目において, 能動的学習を促すことを目的に,プログラミング課題を 解くことを中心に学習する授業の体制を構築し,複数の 教員によるチームティーチングを実践している. この授業ではプログラミング課題を解く過程で生じる 個々の学生からの多様な質問に対応するため,複数の教 員に加えて,スチューデントアシスタント(以後,SAと 記す)が参加し,学習支援を行っている. これまでの授業の取組みでは,プログラミング課題に 取組むための事前学習に対する明確な指示は出しておら ず,事前学習を行うことを学生の自主性に委ねていた. そのため,授業時間に初めて課題内容を把握する学生も 見られ,限られた授業時間内に効果的な指導ができない 状況が見られた. また,授業期間を通して,プログラミングの基本的な 知識や概念の理解が不十分な状況では,その後の発展的 な課題を解くことにも影響を及ぼすため,学習支援が効 果的に行えない状況も見られた. 1.2 研究の目的 上述したこれまでの授業における課題に対して,次の 2つの方針を新たに授業方針に設定した. 1 プログラミング課題に関する事前学習教材を学 生へ提供し,学生は事前学習教材を学習して授業に 参加する. 2 プログラミング課題として基本的な知識や概念 を扱う授業の前半では,事前学習と課題を中心に授 業を進め,後半の授業では,学生の理解度に応じた クラスに再編し,クラスごとに授業を進める. 後半の授業では,前半の授業で指定した課題に関わる 2017 年2月 14 日受付 ※1日本工業大学工学部 事前学習と課題への取り組みを中心とした 初年次プログラミング教育の実践