Machine Learning is the study of methods for programming computers to learn. Computers are applied to a wide range of tasks, and for most of these it is relatively easy for programmers to design and implement the necessary software. However, there are many tasks for which this is difficult or impossible. These can be divided into four general categories. First, there are problems for which there exist no human experts. For example, in modern automated manufacturing facilities, there is a need to predict machine failures before they occur by analyzing sensor readings. Because the machines are new, there are no human experts who can be interviewed by a programmer to provide the knowledge necessary to build a computer system. A machine learning system can study recorded data and subsequent machine failures and learn prediction rules. Second, there are problems where human experts exist, but where they are unable to explain their expertise. This is the case in many perceptual tasks, such as speech recognition, hand-writing recognition, and natural language understanding. Virtually all humans exhibit expert-level abilities on these tasks, but none of them can describe the detailed steps that they follow as they perform them. Fortunately, humans can provide machines with examples of the inputs and correct outputs for these tasks, so machine learning algorithms can learn to map the inputs to the outputs. Third, there are problems where phenomena are changing rapidly. In finance, for example, people would like to predict the future behavior of the stock market, of consumer purchases, or of exchange rates. These behaviors change frequently, so that even if a programmer could construct a good predictive computer program, it would need to be rewritten frequently. A learning program can relieve the programmer of this burden by constantly modifying and tuning a set of learned prediction rules. Fourth, there are applications that need to be customized for each computer user separately. Consider, for example, a program to filter unwanted electronic mail messages. Different users will need different filters. It is unreasonable to expect each user to program his or her own rules, and it is infeasible to provide every user with a software engineer to keep the rules up-to-date. A machine learning system can learn which mail messages the user rejects and maintain the filtering rules automatically. Machine learning addresses many of the same research questions as the fields of statistics, data mining, and psychology, but with differences of emphasis. Statistics focuses on understanding the phenomena that have generated the data, often with the goal of testing different hypotheses about those phenomena. Data mining seeks to find patterns in the data that are understandable by people. Psychological studies of human learning aspire to understand the mechanisms underlying the various learning behaviors exhibited by people (concept learning, skill acquisition, strategy change, etc.).

Machine learning

物件導向軟體之架構(Object-Oriented Software Construction)探討

This paper studies the control of a class of discrete event processes, i.e. processes that are discrete, asynchronous and possibly nondeter-ministic. The controlled process is described as the generator of a formal language, while the controller, or supervisor, is constructed from the grammar of a specified target language that incorporates the desired closed-loop system behavior. The existence problem for a supervisor is reduced to finding the largest controllable language contained in a given legal language. Two examples are provided.

/pdf/supervisory-control-of-a-class-of-discrete-event-processes-45fsmhdhbo.pdf

Supervisory control of a class of discrete event processes

Since the subject of traffic dynamics has captured the interest of physicists, many surprising effects have been revealed and explained. Some of the questions now understood are the following: Why are vehicles sometimes stopped by ``phantom traffic jams'' even though drivers all like to drive fast? What are the mechanisms behind stop-and-go traffic? Why are there several different kinds of congestion, and how are they related? Why do most traffic jams occur considerably before the road capacity is reached? Can a temporary reduction in the volume of traffic cause a lasting traffic jam? Under which conditions can speed limits speed up traffic? Why do pedestrians moving in opposite directions normally organize into lanes, while similar systems ``freeze by heating''? All of these questions have been answered by applying and extending methods from statistical physics and nonlinear dynamics to self-driven many-particle systems. This article considers the empirical data and then reviews the main approaches to modeling pedestrian and vehicle traffic. These include microscopic (particle-based), mesoscopic (gas-kinetic), and macroscopic (fluid-dynamic) models. Attention is also paid to the formulation of a micro-macro link, to aspects of universality, and to other unifying concepts, such as a general modeling framework for self-driven many-particle systems, including spin systems. While the primary focus is upon vehicle and pedestrian traffic, applications to biological or socio-economic systems such as bacterial colonies, flocks of birds, panics, and stock market dynamics are touched upon as well.

Traffic and related self-driven many-particle systems

https://www.ufz.de/export/data/2/100065_GrimmODD.pdf

A standard protocol for describing individual-based and agent-based models

Systems Theory Challenges in the Simulation of Variable Structure and Intelligent Systems

In the Winter Simulation Conference 1999 in Phoenix, a series of discussions, conversations, and exchanges on the topic of personnel to meet the current modeling and simulation demands of the civilian application and world military as well led to the idea of giving this discussion a more structured shape in the Winter Simulation Conference 2000. We continue to discuss the complex issue of Simulation Education. A general demand for modeling and simulation professionals can be observed in a large number of enterprises. However computer science graduates are not adequately prepared for employment opportunities involving simulation as a tool in solving problems. Most computer science majors have very limited exposure to simulation. They gain experience in handling of simulation problems by on-the-job-training. Moreover, there doesn't exist any consensus of simulation as a discipline. The following questions hence emerge:• What are the reasons for shortages of modeling and simulation professionals?• What would make simulation into a discipline?• What skills should professionals develop during the education and training?• Impact of developments in simulation technology: what do we educate simulation professionals for?• What are the educational strategies to meet current and anticipated world needs in simulation?• What are the goals of an educational curriculum for simulation?• How to organize education of simulation to make it attractive for students?• What are criteria on selection of tools for teaching simulation?• Are there initiatives currently going on in the Modelling and Simulation community to establish some structure in the M&S education and training?The panel collects 6 simulation professionals from educational institutions that currently offer simulation programs, and non-educational organizations with interests in simulation education. The objective is to address issues related to the growth and need of degree programs in simulation. The panel members from academia, enterprises when the outcome is not known but is sought to aid in the creation of new knowledge either for its own sake or aid decision maker in a decision process. Entertainment simulation provides an experience where the outcome is simply entertainment While these three categories are useful, there are not sharp boundaries between the three. There is overlap among all three categories. Simulation environments can be used for all three simultaneously. At Old Dominion, the focus is on environments for training and education and, especially, on environments for discovery. Of particular interest are those instances where outcomes of the simulation environment include both training/education and discovery.

Conceptions of curriculum for simulation education: panel

Powered by the rapid advance of computer, network, and sensor/actuator technologies, distributed real-time systems that continually and autonomously control and react to the environment have been widely used. The combination of temporal requirements, concurrent environmental entities, and high reliability requirements, together with distributed processing make the software to control these systems extremely hard to design and difficult to verify. 
In this work, we developed a simulation-based software development methodology to manage the complexity of distributed real-time software. This methodology, based on discrete event system specification (DEVS), overcomes the “incoherence problem” between different design stages by emphasizing “model continuity” through the development process. Specifically, techniques have been developed so that the same control models that are designed can be tested and analyzed by simulation methods and then easily deployed to the distributed target system for execution. To improve the traditional software testing process where real-time embedded software needs to be hooked up with real sensor/actuators and placed in a physical environment for meaningful test and analysis, we developed a virtual test environment that allows software to be effectively tested and analyzed in a virtual environment, using virtual sensor/actuators. Within this environment, stepwise simulation methods have been developed so that different aspects, such as logic and temporal behaviors, of a real-time system can be tested and analyzed incrementally. 
Based on this methodology, a simulation and testing environment for distributed autonomous robotic systems is developed. This environment has successfully supported the development and investigation of several distributed autonomous robotic systems. One of them is a “dynamic team formation” system in which mobile robots search for each other, and then form a team dynamically through self-organization. Another system is a scalable robot convoy system in which robots convoy and maintain a line formation in a coordinated way.

/pdf/a-simulation-based-software-development-methodology-for-2ece7bh4eb.pdf

A simulation-based software development methodology for distributed real-time systems

From a modeling and simulation perspective, studying dynamic systems consists of focusing on changes in states. According to the precision of state changes, generic algorithms can be developed to track the activity of sub-systems. This paper aims at describing and applying this more natural and intuitive way to describe and implement dynamic systems. Activity is defined mathematically. A generic application case of diffusion is experimented with to compare the efficiency of quantized state methods using this new approach with traditional methods which do not focus computations on active areas. Our goal is to demonstrate that the concept of activity can estimate the computational effort required by a quantized state method. Specifically, when properly designed, a discrete-event simulator for such a method achieves a reduction in the number of state transitions that more than compensates for the overhead it imposes.

/pdf/the-activity-tracking-paradigm-in-discrete-event-modeling-41fs11lv52.pdf

The Activity-tracking paradigm in discrete-event modeling and simulation: The case of spatially continuous distributed systems

A healthcare service system (HSS) is made of humans and technology where for the foreseeable future, self-improvement will be primarily based on human understanding rather than machine learning. Therefore, for such a system to continually self-improve, it must provide the right data and models to support human decisions on selection of alternatives likely to improve the quality of its services. Our focus in this paper is to show how modeling and simulation (M&S) can help design service infrastructures that introduce coordination and bring into play the conditions for learning and continuous improvement. To do this, we discuss the application of the discrete event system specification (DEVS) formalism within system of systems engineering (SoSE) to develop coordination models for transactions that involve multiple disparate activities of component systems and that need to be selectively sequenced to implement patient-centered coordinated care interventions. We show how such coordination concepts provide a layer to support a proposed information technology for continuous improvement of healthcare as a learning collaborative system of systems (SoS).

Bernard P. Zeigler

Papers

Systems Theory Challenges in the Simulation of Variable Structure and Intelligent Systems

Conceptions of curriculum for simulation education: panel

A simulation-based software development methodology for distributed real-time systems

The Activity-tracking paradigm in discrete-event modeling and simulation: The case of spatially continuous distributed systems

Discrete Event System Specification Framework for Self-Improving Healthcare Service Systems