Michael North
Vice President, Decision Sciences, Conversant LLC (USA)

More Information

Jiming Liu
Chair Professor of Computer Science, Hong Kong Baptist University

More Information

Petr Skobelev
Smart Solutions / Samara State Airspace and Technical Universities / Institute of Control of Complex Systems of Russian Academy of Science

More Information

Michael

Michael North

Vice President, Decision Sciences, Conversant LLC (USA)

Title: Hammer or Tongs: How Best to Build Agent-Based Models?
Abstract

Agent-based modeling has been widely applied to solve problems in contexts ranging from exploratory research studies to focused industrial practice. The methodologies employed in these efforts to develop and use agent-based models have run the gamut from ad hoc coding to formal methods. This talk will survey the existing methodologies for developing and using agent-based models; consider their relative strengths and weaknesses; and offer insights into how to choose one for your next agent-based modeling project.

Biography
Jiming

Jiming Liu

Chair Professor of Computer Science, Hong Kong Baptist University

Title: On Data-Driven Agent-Based Modeling of Complex Systems
Abstract

Complex systems modeling plays a pivotal role in characterizing and understanding biological, ecological, social, and technological systems. This talk will discuss several important issues in developing agent-based complex systems modeling. Specifically, the talk will address the promises and challenges of data-driven agent-based modeling in unveiling hidden mechanisms as well as intrinsic behavior of complex systems. The talk will look into several examples in the fields of computational healthcare and sustainability.

Biography
Petr Skobelev

Petr Skobele

Smart Solutions / Samara State Airspace and Technical Universities / Institute of Control of Complex Systems of Russian Academy of Science

Title: Towards Smart Enterprise 4.0/5.0: Knowledge bases and multi-agent technology for real time adaptive scheduling - industrial applications
Abstract

Key business requirements and complexity of real time resource management for industrial applications, working under conditions of high uncertainty and dynamics of unpredictable events, will be analyzed. It will be shown that complexity of resource management in business is related not only with number of orders and resources and combinatorial NP-hard search of variants in solution space but also with number of decision makers with conflicting interests, high variety of orders and resources, long list of factors, individual criteria, preferences and constraints for orders and resources, interdependency of all operations, etc. The concept of real time adaptive scheduling on virtual market of agents will be presented and theoretical framework of multi-iterative auctions will be overviewed. Developed models and methods for solving conflicts and finding agents consensus in coordinated multi-criteria decision making will be discussed and the analogue with not-linear thermodynamics and unstable equilibriums will be shown. Ontology for scheduling will be introduced and how it helps to build ontological model of the enterprise and customize matching requirements for each operation in business or technological processes. Semantic Wikipedia on the top of ontology editor will be discussed which helps to build knowledge base of enterprise for resource management. The way how to create and measure emergent “adaptive intelligence” will be demonstrated for smart swarm of satellites which is able to self-organize and adapt dynamically to unpredictable events. Business experience and results of delivery of adaptive multi-agent solutions for trucks and factories, mobile teams, supply chains, aerospace and railways will be presented. As a main result it will be shown that multi-agent technology helps to increase efficiency of enterprise resources up to 15-40% and what kind of theoretical and practical difficulties need to be overcomed. Future trends on Smart Enterprise 4.0 / 5.0 will be outlined and work in progress on self-organized “system of systems” for solving extremely complex problems of real time scheduling will be discussed.

Biography