Corporate dashboards for integrated business and engineering decisions in oil refineries: An agent-based approach

It is generally very challenging for an oil refinery to make integrated decisions encompassing multiple functions based on a traditional Decision Support System (DSS), given the complexity and interactions of various decisions. To overcome this limitation, we propose an integrated DSS framework by combining both business and engineering systems with a dashboard. The dashboard serves as a human-computer interface and allows a decision maker to adjust decision variables and exchange information with the DSS. The proposed framework provides a two-stage decision making mechanism based on optimization and agent-based models. Under the proposed DSS, the decision maker decides on the values of a subset of decision variables. These values, or the first-stage decision, are forwarded through the dashboard to the DSS. For the given set of first-stage decision variables, a multi-objective robust optimization problem, based on an integrated business and engineering simulation model, is solved to obtain the values for a set of second-stage decision variables. The two-stage decision making process iterates until a convergence is achieved. A simple oil refinery case study with an example dashboard demonstrates the applicability of the integrated DSS.     

面向深度学习的公平性研究综述

深度学习是机器学习研究中的一个重要领域,它具有强大的特征提取能力,且在许多应用中表现出先进的性能,因此在工业界中被广泛应用.然而,由于训练数据标注和模型设计存在偏见,现有的研究表明深度学习在某些应用中可能会强化人类的偏见和歧视,导致决策过程中的不公平现象产生,从而对个人和社会产生潜在的负面影响.为提高深度学习的应用可靠性、推动其在公平领域的发展,针对已有的研究工作,从数据和模型2方面出发,综述了深度学习应用中的偏见来源、针对不同类型偏见的去偏方法、评估去偏效果的公平性评价指标、以及目前主流的去偏平台,最后总结现有公平性研究领域存在的开放问题以及未来的发展趋势.     

基于多智能体的人机协作智能信息系统

综合集成研讨厅理论体系是作为处理开放的复杂巨系统的有关问题而提出的,其实质是一个人机协作的智能信息系统.本文主要探索研讨厅的实现途径,给出了基于网络的研讨厅的层次模型与系统结构,提出了基于内嵌客户/Agent/服务器的请求器-中介器-供应器模型的多智能体研讨厅的多层分布计算模型,并基于Java的智能信息Agent技术探讨了支持宏观经济决策的多智能体研讨厅的构架及工作机制.研究表明,基于Internet有关技术与Agent计算技术有可能实现研讨厅,并由于Agent技术的优势而可能得到较好的系统性能.     

Towards a Framework for Visual Model Development

A number of decision support systems (DSS) based on a visual interactive simulation (VIS) model have been implemented in recent years. A key component of a VIS model is the visual model. Existing VIS model development methodologies suggest methods (such as the paper prototype method) for developing a visual model. However, these methods are quite general and do not contain a framework to guide development. This paper describes a framework which combines the paper prototype method with systems concepts to develop the visual model. The proposed framework consists of eight stages. A project proof of principle example is described in which a VIS-based DSS provides support to the owner of a company selling telephone systems.     

基于多Agent的电力营销决策支持系统的研究

针对传统决策支持系统的局限性以及决策支持问题的复杂性,本文把Agent方法引入到决策支持系统的研究中,提出了一个新型的决策支持系统框架,设计了基于多Agent的电力营销决策支持系统。该模型中的多个Agent通过Agent解释服务通信交流,相互合作,共同完成决策支持任务,充分的发挥了Agent的自主性、反应性、协作性,为电力企业的决策问题的解决提供了一个新的途径。     

The Importance of Active Exploration,Optical Flow,and Task Alignment for Spatial Learning in Desktop 3D Environments

The Decision Support Systems (DSS) field has grown rapidly, drawing technology from many disciplines and pursuing applications in a variety of domains but developing little underlying theoretical structure and poor linkage between research and practice. This article presents a classification scheme for DSS techniques that provides a common theoretical framework for DSS research and also structures and simplifies the process of designing application systems. The classification system is functional, grouping DSS techniques according to their ability to provide similar kinds of support (i.e., functions) to a human decision maker. It is also cognitively based, defining the kinds of support that decision makers need in terms of architectural features and procedural aspects of human cognition. The classification is expressed as a taxonomy, encompassing six primary classes of decision support techniques representing the six general kinds of cognitive support that human decision makers need. The six classes are process models, which assist in projecting the future course of complex processes; choice models, which support integration of decision criteria across aspects and/or alternatives; information control techniques, which help in storage, retrieval, organization, and integration of data and knowledge; analysis and reasoning techniques, which support application of problem-specific expert reasoning procedures; representation aids, which assist in expression and manipulation of a specific representation of a decision problem; and judgment amplification/refinement techniques, which help in quantification and debiasing of heuristic judgments. Additional distinctions are provided to distinguish the individual techniques in each of these primary categories. The taxonomy also has practical use as a design aid for decision support systems. The kinds of decision support needs represented by the taxonomy are general and can be used to guide the analysis and decomposition of a given decision prior to decision aid design. Specific needs for assistance can then be tied to specific computational techniques in the taxonomy. Methodological suggestions for using the taxonomy as a design aid are given.     

Debiasing training data for inductive expert system construction

We study the presence of economic bias in the training data used to develop inductive expert systems. Such bias arises when an expert considers economic factors in decision making. We find that the presence of economic bias is particularly harmful when there is an economic misalignment between the expert and the user of the induced expert system. Such misalignment is referred to as differential bias. The most significant contribution of this study is a training data debiasing procedure that uses a genetic algorithm to reconstruct training data that is relatively free of economic bias. We conduct a series of simulation experiments that show: the economic performance of accuracy and value seeking algorithms is statistically the same when the training data has economic bias; both accuracy and value seeking algorithms suffer in the presence of differential bias; the proposed debiasing procedure significantly combats differential bias; and the debiasing procedure is quite robust with respect to estimation errors in its input parameters     

Cognitive biases and decision support systems development: a design science approach

Abstract.  This paper presents design science research that aims to improve decision support systems (DSS) development in organizations. Evolutionary development has been central to DSS theory and practice for decades, but a significant problem for DSS analysts remains how to conceptualize the improvement of a decision task during evolutionary DSS development. The objective of a DSS project is to improve the decision process and outcome for a manager making an important decision. The DSS analyst needs to have a clear idea of the nature of the target decision task and a clear strategy of how to support the decision process. Existing psychological research was examined for help with the conceptualization problem, and the theory of cognitive bias is proposed as a candidate for this assistance. A taxonomy of 37 cognitive biases that codifies a complex area of psychological research is developed. The core of the project involves the construction of a design artefact – an evolutionary DSS development methodology that uses cognitive bias theory as a focusing construct, especially in its analysis cycles. The methodology is the major contribution of the project. The feasibility and effectiveness of the development methodology are evaluated in a participatory case study of a strategic DSS project where a managing director is supported in a decision about whether to close a division of a company.     

An agent-based system for orchestration support of web service-enabled devices in discrete manufacturing systems

New trends in the automation industry indicate that manufacturing companies can improve their response to the constantly changing market demands for customized products by adopting the design principles from the Service-Oriented Architecture (SOA) in their control and automation systems. With this goal in mind, the EU FP6 IST project Service-Oriented Cross-layer infRAstructure for Distributed smart Embedded devices (SOCRADES), researched the creation of manufacturing systems with distributed control architectures, where Web Services (WS)s are pushed all the way down to the device level. The results of this research allowed us to expose the functionality of production equipment by means of WSs and use them according to the SOA principles; however, it was found that it is necessary to count with a mechanism for ensuring smart and dynamic composition of these WSs on the factory floor. It is presumed that agent-based technology has the potential of providing the required level of intelligence, if it can be properly combined with WSs on manufacturing systems. This article describes the specification of a WS-enabled Decision Support System (DSS) integrated by a set of software agents. The agent-based system presented here is capable of supporting the dynamic composition and orchestration of WSs exposed by control devices on discrete manufacturing systems. In addition to the DSS specification, this paper also introduces a physical scenario which was used to test the proposed system. The developments and results presented here were also done within the framework of the SOCRADES project.     

Decision support systems: An expert systems approach

This paper provides a conceptual framework for designing decision support systems (DSS) using an expert systems approach. Currently there is a significant trend towards the use of knowledge-based systems techniques in DSS design, but a comprehensive framework is yet to be proposed. Our paper addresses this problem and presents such a framework. Efforts are currently underway to design, implement and test a system based on this framework.     

Bias Reduction in Variational Regularization

The aim of this paper was to introduce and study a two-step debiasing method for variational regularization. After solving the standard variational problem, the key idea is to add a consecutive debiasing step minimizing the data fidelity on an appropriate set, the so-called model manifold. The latter is defined by Bregman distances or infimal convolutions thereof, using the (uniquely defined) subgradient appearing in the optimality condition of the variational method. For particular settings, such as anisotropic \(\ell ^1\) and TV-type regularization, previously used debiasing techniques are shown to be special cases. The proposed approach is, however, easily applicable to a wider range of regularizations. The two-step debiasing is shown to be well-defined and to optimally reduce bias in a certain setting. In addition to visual and PSNR-based evaluations, different notions of bias and variance decompositions are investigated in numerical studies. The improvements offered by the proposed scheme are demonstrated, and its performance is shown to be comparable to optimal results obtained with Bregman iterations.     

Declarative Bias for Specific-to-General ILP Systems

A comparative study is presented of language biases employed in specific-to-general learning systems within the Inductive Logic Programming (ILP) paradigm. More specifically, we focus on the biases employed in three well known systems: CLINT, GOLEM and ITOU, and evaluate both conceptually and empirically their strengths and weaknesses. The evaluation is carried out within the generic framework of the NINA system, in which bias is a parameter. Two different types of biases are considered: syntactic bias, which defines the set of well-formed clauses, and semantic bias, which imposes restrictions on the behaviour of hypotheses or clauses. NINA is also able to shift its bias (within a predefined series of biases), whenever its current bias is insufficient for finding complete and consistent concept definitions. Furthermore, a new formalism for specifying the syntactic bias of inductive logic programming systems is introduced.This paper extends the papers (Adé & Bruynooghe, 1992) and (Rouveirol et al., 1993).     

Pluralistic multi-agent decision support system: a framework and an empirical test

Recent research in decision support systems (DSSs) has focused on building active cooperative intelligent systems. Research in agent-based decision support is a promising stream in this direction. This paper proposes a framework for a pluralistic multi-agent decision support system (MADSS). The distinguishing feature of the proposed approach is its organization around human decision making process. The framework builds upon the decision support pyramid with agents organized into groups according to the phases of the problem solving model. We outline the design principles and develop architecture for MADSS. The framework is illustrated through an investment MADSS prototype. The results of the empirical test are presented.     

Evaluation and Selection of Biases in Machine Learning

In this introduction, we define the termbias as it is used in machine learning systems. We motivate the importance of automated methods for evaluating and selecting biases using a framework of bias selection as search in bias and meta-bias spaces. Recent research in the field of machine learning bias is summarized.     

An intelligent agent-based architecture for strategic information system applications

Strategic information systems (SIS) focus on the use of information system (IS) and information technology (IT) in the strategic management process in business organizations. The emphasis is on the strategic view of IS and IT and their impact on organizational strategy. Increased competition and advances in information technologies push for considerable structural changes in SIS. Agents, as autonomous entities which either work on their own or cooperate with others, and agent architectures have enormous potentials to be applied in such critical systems. In this article, first we investigate the very fundamental concepts of strategic information systems and intelligent agent technology. Then, the discussion continues on the specification of the characteristics and implementation issues of a typical SIS. Afterwards, we make use of these concepts and integrate them into a state-of-the-art, intelligent architecture for strategic information systems, called intelligent agent-based SIS. This is a comprehensive framework for a SIS in IT era which may be put into practice by a team of professionals in the near future. The graphical representation of this model is intended to help the reader understand the concept much better. After explaining the suggested model in full details, we introduce some support agents and specify their corresponding roles in an intelligent agent-based SIS architecture. Discussions and concluding remarks regarding the proposed system are provided at the end of the paper.     

A Jini‐based software framework for developing distributed cooperative decision support systems

Many mission‐critical, decision‐making situations happen in dynamic, rapidly changing, and often unpredictable environments. Military, governmental, and medical contexts are examples of such situations, which can be characterized by highly decentralized, up‐to‐date data sets coming from various sources. Unlike other decision‐making tools, decision support systems (DSS) designed for such situations are challenged by the need to access this decentralized data at any time, from anywhere, under tight time constraints. This paper presents the design of a software framework for developing these kinds of distributed DSSs. The proposed Java‐based framework relies mostly on the Jini technology and its JavaSpaces service. The construction of the framework is presented with UML class diagrams, UML sequence diagrams, and additional explanations. Copyright © 2003 John Wiley & Sons, Ltd.     

Agent-based framework for integrated facility engineering

An agent-based framework for the development of integrated facility engineering environments in support of collaborative design is introduced. This framework aims at integrating design software by allowing better software interoperability. Within their framework, design agents represent various existing design and planning systems that communicate their design information and knowledge partially and incrementally using the Agent Communication Language (ACL). ACL is a formal language proposed as a communication standard for disparate software. It is based on a logic-based language called Knowledge Interchange Format (KIF) and a message protocol called Knowledge Query Manipulation Language (KQML). Design agents are linked and their communication of design information is coordinated via system programs called facilitators in a federation architecture. The federation architecture specifies the way design agents and facilitators communicate in an integrated software environment. In concert with pursuing fundamental research concepts, we have been developing an integrated design software environment that spans different phases of the facility life cycle. This environment serves to demonstrate the primary aspects of this research methodology. In this paper, we first discuss the integration problem and review related research projects. We then present the major aspects of agent-based software engineering methodology and its application to integrated facility engineering. A highlight of the current integrated design environment development is given to illustrate the advantages of this approach. Finally, we summarize and discuss some of the important research issues in light of previous research.     

分布式决策中处理冲突的研究

在基于agent的DSS系统中，由于不同的agent具有不同的评价标准，因而必然存在冲突．本文在形式化评价决策任务的基础上，提出了通过利用评价agent改变当前属性位并把有关结果通知其它协商agent来解决冲突的方案，并给出了消除冲突的具体过程．最后，通过具体实例说明了该方法的可行性。     

Rule fitness and pathology in learning classifier systems

It has long been known that in some relatively simple reinforcement learning tasks traditional strength-based classifier systems will adapt poorly and show poor generalisation. In contrast, the more recent accuracy-based XCS, appears both to adapt and generalise well. In this work, we attribute the difference to what we call strong over general and fit over general rules. We begin by developing a taxonomy of rule types and considering the conditions under which they may occur. In order to do so an extreme simplification of the classifier system is made, which forces us toward qualitative rather than quantitative analysis. We begin with the basics, considering definitions for correct and incorrect actions, and then correct, incorrect, and overgeneral rules for both strength and accuracy-based fitness. The concept of strong overgeneral rules, which we claim are the Achilles' heel of strength-based classifier systems, are then analysed. It is shown that strong overgenerals depend on what we call biases in the reward function (or, in sequential tasks, the value function). We distinguish between strong and fit overgeneral rules, and show that although strong overgenerals are fit in a strength-based system called SB-XCS, they are not in XCS. Next we show how to design fit overgeneral rules for XCS (but not SB-XCS), by introducing biases in the variance of the reward function, and thus that each system has its own weakness. Finally, we give some consideration to the prevalence of reward and variance function bias, and note that non-trivial sequential tasks have highly biased value functions.     

A systematic approach to evaluating domain-specific modeling language environments for multi-agent systems

Multi-agent systems (MASs) include multiple interacting agents within an environment to provide a solution for complex systems that cannot be easily solved with individual agents or monolithic systems. However, the development of MASs is not trivial due to the various agent properties such as autonomy, responsiveness, and proactiveness, and the need for realization of the many different agent interactions. To support the development of MASs various domain-specific modeling languages (DSMLs) have been introduced that provide a declarative approach for modeling and supporting the generation of agent-based systems. To be effective, the proposed DSMLs need to meet the various stakeholder concerns and the related quality criteria for the corresponding MASs. Unfortunately, very often the evaluation of the DSML is completely missing or has been carried out in idiosyncratic approach. If the DSMLs are not well defined, then implicitly this will have an impact on the quality of the MASs. In this paper, we present an evaluation framework and systematic approach for assessing existing or newly defined DSMLs for MASs. The evaluation is specific for MAS DSMLs and targets both the language and the corresponding tools. To illustrate the evaluation approach, we first present SEA_ML, which is a model-driven MAS DSML for supporting the modeling and generation of agent-based systems. The evaluation of SEA_ML is based on a multi-case study research approach and provides both qualitative evaluation and quantitative analysis. We report on the lessons learned considering the adoption of the evaluation approach as well as the SEA_ML for supporting the generation of agent-based systems.