Rustam M. Vahidov

Application of machine learning techniques for supply chain demand forecasting

Abstract Full collaboration in supply chains is an ideal that the participant firms should try to achieve. However, a number of factors hamper real progress in this direction. Therefore, there is a need for forecasting demand by the participants in the absence of full information about other participants’ demand. In this paper we investigate the applicability of advanced machine learning techniques, including neural networks, recurrent neural networks, and support vector machines, to forecasting distorted demand at the end of a supply chain (bullwhip effect). We compare these methods with other, more traditional ones, including naive forecasting, trend, moving average, and linear regression. We use two data sets for our experiments: one obtained from the simulated supply chain, and another one from actual Canadian Foundries orders. Our findings suggest that while recurrent neural networks and support vector machines show the best performance, their forecasting accuracy was not statistically significantly better than that of the regression model.

Decision station: situating decision support systems

Internet facilitates access to data, information, and knowledge sources, but at the same time, it threatens to cognitively overload the decision makers. This necessitates the development of effective decision support tools to properly inform the decision process. Internet technologies require new type of decision support that provides tighter integration and higher degree of direct interaction with the problem domain. The central argument of this work is that in dynamic and highly complex electronic environments decision support systems (DSSs) should be situated in the problem domain. A generic architecture, the set of capabilities for our vision of a situated DSS is proposed, and the architecture is illustrated with a DSS for investment management.

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.

e-Negotiation Systems and Software Agents: Methods, Models, and Applications

Negotiation is a decentralized decision-making process that seeks to find an agreement that will satisfy the requirements of two or more parties in the presence of limited common knowledge and conflicting preferences. Negotiation participants are agents who negotiate on their own behalf or represent the interests of their principals. When electronic negotiations enter the stage, these agents could be intelligent software entities that take part in the process of searching for an acceptable agreement. The degree of involvement of these “intelligent agents” in negotiations can range from supporting human negotiators (e. g. information search, offer evaluation) to fully automating the conduct of negotiations. Choosing the degree of involvement depends upon the characteristics of the problem in the negotiation. In this chapter, we review electronic negotiation systems and intelligent agents for negotiations. Different types of negotiation agents, their roles and requirements, and various methods for effective support or conduct of negotiations are discussed. Selected applications of intelligent negotiation agents are presented.

Agent-supported negotiations in the e-marketplace

A key requirement for customer-oriented e-marketplaces is the personalisation and customisation of products and services, as well as personalised fulfilment. Negotiations are at the core of these processes. In this paper an e-marketplace, eAgora, is discussed. It allows buyers and sellers to engage in multi-issue negotiations. eAgora implements several protocols based on a negotiation phase model constructed for this system. Its services include a software agent that generates and critiques offers. Based on a small-scale usability testing with participants, who conducted negotiations with and without the agent, the agents services were found useful. The users also requested additional agents services including enhanced offer critique and partial negotiation automation.

Load Shifting Agents for Automated Demand Side Management in Micro Energy Grids

This paper describes a novel approach for the automated management of micro energy grids. In particular a market based resource allocation mechanism is used to control energy generators and consumers within a micro energy grid. This approach requires energy consumers (producers) to buy (sell) their energy demands (supplies) through a specialized electronic auction platform. But as manually negotiating all energy demands and supplies on such a market is a tedious task, its automation is highly desirable and thus leads to the main contribution of this paper: The automation of the demand side bidding process through electronic bidding agents, which are equipped with an intelligent buying strategy that allows them to dynamically react to market changes and adapt their bidding behavior accordingly. More precisely, the agents are able to shift energy demand within certain boundaries from (expensive) peak hours to those times of the day where energy demand and thus energy prices are lower in order to minimize their cost. Moreover, as our results show, this behavior leads to a smoothed load curve for the whole system, i.e. demand peaks are reduced while base load increases.

An experimental study of software agent negotiations with humans

Electronic negotiations allow participants to negotiate online and use analytical support tools in making their decisions. Software agents offer the possibility of automating negotiation process using these tools. This paper aims at investigating the prospects of agent-to-human negotiations using experiments with human subjects. Various types of agents have been configured using the following tactics: individualistic, neutral, yielding, yielding-then-individualistic, and absolute tit-for-tat. These agents were paired up with human counterparts for negotiating product sale. A set of hypotheses has been proposed involving the performance of agents, as well as humans in terms of objective, as well as subjective measures. Overall, the findings speak in favor of agent-managed negotiations. An experiment in software agent - human negotiations is described.Agents employed different negotiation tactics.The negotiation case involving purchase of a computer was used.Subjects were university students.Overall, software agents outperformed human negotiators.

Concession-making in multi-attribute auctions and multi-bilateral negotiations: Theory and experiments

Concession-making behavior is an essential process in negotiations and auctions and has critical impact on the outcomes of an exchange. In auctions, concessions relate to deciding on the next bid by bid-makers, while in negotiations they involve proposing next offers by both parties. The purpose of this paper is twofold: (1) present a theory of concessions which could be applied to both auctions and negotiations and (2) provide experimental verification of the theory. The concession model identifies nine types of concessions derived from the preference structures. The occurrence of all nine types has been confirmed in experiments. The model also identifies two concession categories which are associated with the process transparency. The theory-based assessment of the concession-making in multi-attribute auctions and multi-issue multi-bilateral negotiations allows for their comparison. To this end, two systems implemented on the Invite e-negotiation platform are used. One of the major findings suggests that auctions induce concession-making that is more likely to lead to Pareto-optimal agreements. This study also suggests that greater transparency in negotiations is likely to lead to better agreements.

Pairwise issue modeling for negotiation counteroffer prediction using neural networks

Electronic negotiation systems can incorporate computational models and algorithms in order to help negotiators achieve their objectives. An important opportunity in this respect is the development of a component, which can assess an expected reaction by a counterpart to a given trial offer before it is submitted. This work proposes a pairwise modeling approach that provides the possibility of developing flexible and generic models for counteroffer prediction when the negotiation cases are similar. The key feature is that each negotiated issue is predicted while paired with each of the other issues and the permutations of issue pairs across all negotiation offers are confounded together. This data fusion permits extractions of common relationships across all issues, resulting in a type of pattern fusion. Experiments with electronic negotiation data demonstrated that the models predictive performance is equivalent to case-specific models while offering a high degree of flexibility and generality even when predicting to a new issue.

On Comparison of Mechanisms of Economic and Social Exchanges: The Times Model

An e-market system is a concrete implementation of a market institution; it embeds one or more exchange mechanisms. E-market systems are also information systems which are information and communication technologies artifacts. This work puts forward an argument that the study of e-markets must incorporate both the behavioral economics as well as the information systems perspectives. To this end the paper proposes a conceptual framework that integrates the two. This framework is used to formulate a model, which incorporates the essential features of exchange mechanisms, as well as their implementations as is artefacts. The focus of attention is on two classes of mechanisms, namely auctions and negotiations. They both may serve the same purpose and their various types have been embedded in many e-market systems.