Utility distribution matters: enabling fast belief propagation for multi-agent optimization with dense local utility function

2021 ◽  
Vol 35 (2) ◽  
Author(s):  
Yanchen Deng ◽  
Bo An
Keyword(s):  
Utility Function ◽  
Belief Propagation ◽  
Multi Agent ◽  
Utility Distribution

scholarly journals FAULT-TOLERANT MULTI-AGENT EXACT BELIEF PROPAGATION

2009 ◽  
Vol 25 (1) ◽  
pp. 1-30 ◽  
Author(s):  
Xiangdong An ◽  
Nick Cercone
Keyword(s):  
Belief Propagation ◽  
Fault Tolerant ◽  
Multi Agent

Simulation of Skulduggery in a Multi-Agent System

2015 ◽  
Vol 7 (1) ◽  
pp. 17-31
Author(s):  
Rasoul Ramezanian ◽  
Akram Emdadi
Keyword(s):  
Utility Function ◽  
Test Session ◽  
Testing Session ◽  
Multi Agent System ◽  
Agent System ◽  
Factors Affecting ◽  
Artificial Society ◽  
Control Factors ◽  
Multi Agent ◽  
Level Of Knowledge

In a testing session, students may want to use the information of other students, which is cheating. The authors of this paper develop an artificial society to model and simulate this situation. They consider two control factors to increase the incentive of students to not cheat. The first factor is the penalty for similarity between responses (as much as two answer-sheets of two students are the same, their final grades decrease). The second factor is the observers who look into the students and do not allow the observed students to cheat. In this model, agents participate in a test based on their level of knowledge, location and two above factors, deciding whether or not to cheat. These components are used to formulate the utility function. Taking advantage of the developed artificial society, the authors now study the above factors affecting the amount of cheating in a test session.

A Simulation of Temporally Variant Agent Interaction via Belief Promulgation

2012 ◽  
pp. 913-927
Author(s):  
Adam J. Conover
Keyword(s):  
Message Passing ◽  
Belief Propagation ◽  
Autonomous Agent ◽  
Emergent Properties ◽  
Agent Behavior ◽  
Temporal Variance ◽  
Agent Interaction ◽  
Multi Agent ◽  
Asynchronous Environments ◽  
Global Population

This chapter concludes a two part series which examines the emergent properties of multi-agent communication in “temporally asynchronous” environments. Many traditional agent and swarm simulation environments divide time into discrete “ticks” where all entity behavior is synchronized to a master “world clock”. In other words, all agent behavior is governed by a single timer where all agents act and interact within deterministic time intervals. This discrete timing mechanism produces a somewhat restricted and artificial model of autonomous agent interaction. In addition to the behavioral autonomy normally associated with agents, simulated agents should also have “temporal autonomy” in order to interact realistically. This chapter focuses on the exploration of a grid of specially embedded, message-passing agents, where each message represents the communication of a core “belief”. Here, we focus our attention on the how the temporal variance of belief propagation from individual agents induces emergent and dynamic effects on a global population.

scholarly journals Speeding Up Incomplete GDL-based Algorithms for Multi-agent Optimization with Dense Local Utilities

2020 ◽  
Author(s):  
Yanchen Deng ◽  
Bo An
Keyword(s):  
Utility Function ◽  
Branch And Bound ◽  
State Of The Art ◽  
Search Space ◽  
Computational Effort ◽  
The State ◽  
Computational Overhead ◽  
Multi Agent

Incomplete GDL-based algorithms including Max-sum and its variants are important methods for multi-agent optimization. However, they face a significant scalability challenge as the computational overhead grows exponentially with respect to the arity of each utility function. Generic Domain Pruning (GDP) technique reduces the computational effort by performing a one-shot pruning to filter out suboptimal entries. Unfortunately, GDP could perform poorly when dealing with dense local utilities and ties which widely exist in many domains. In this paper, we present several novel sorting-based acceleration algorithms by alleviating the effect of densely distributed local utilities. Specifically, instead of one-shot pruning in GDP, we propose to integrate both search and pruning to iteratively reduce the search space. Besides, we cope with the utility ties by organizing the search space of tied utilities into AND/OR trees to enable branch-and-bound. Finally, we propose a discretization mechanism to offer a tradeoff between the reconstruction overhead and the pruning efficiency. We demonstrate the superiorities of our algorithms over the state-of-the-art from both theoretical and experimental perspectives.

INNOVATION AND SELF-ORGANIZATION IN A MULTI-AGENT MODEL

2009 ◽  
Vol 12 (02) ◽  
pp. 233-253 ◽  
Author(s):  
TANYA ARAÚJO ◽  
R. VILELA MENDES
Keyword(s):  
Utility Function ◽  
The Other ◽  
Self Organization ◽  
Abstract Model ◽  
Agent Based ◽  
Agent Model ◽  
Multi Agent

A model is developed to study the effectiveness of innovation and its impact on structure creation on agent-based societies. The abstract model that is developed is easily adapted to any particular field. In an interacting environment, the agents receive something from the environment (the other agents) in exchange for their effort and pay the environment a certain amount of value for the fulfilling of their needs or for the very price of existence in that environment. This is coded by two bit strings and the dynamics of the exchange is based on the matching of these strings to those of the other agents. Innovation is related to the adaptation by the agents of their bit strings to improve some utility function.

A Simulation of Temporally Variant Agent Interaction via Belief Promulgation

2009 ◽  
pp. 261-274
Author(s):  
Adam J. Conover
Keyword(s):  
Message Passing ◽  
Belief Propagation ◽  
Autonomous Agent ◽  
Emergent Properties ◽  
Agent Behavior ◽  
Temporal Variance ◽  
Agent Interaction ◽  
Multi Agent ◽  
Asynchronous Environments ◽  
Global Population

This chapter concludes a two part series which examines the emergent properties of multi-agent communication in “temporally asynchronous” environments. Many traditional agent and swarm simulation environments divide time into discrete “ticks” where all entity behavior is synchronized to a master “world clock”. In other words, all agent behavior is governed by a single timer where all agents act and interact within deterministic time intervals. This discrete timing mechanism produces a somewhat restricted and artificial model of autonomous agent interaction. In addition to the behavioral autonomy normally associated with agents, simulated agents should also have “temporal autonomy” in order to interact realistically. This chapter focuses on the exploration of a grid of specially embedded, message-passing agents, where each message represents the communication of a core “belief”. Here, we focus our attention on the how the temporal variance of belief propagation from individual agents induces emergent and dynamic effects on a global population.

Reconciling Co-Evolving Engineering and Customer Requirements With a Looped Bayesian Model

2019 ◽  
Author(s):  
Christopher Slon ◽  
Vijitashwa Pandey
Keyword(s):  
Utility Function ◽  
Real Life ◽  
Decision Makers ◽  
Shape Parameters ◽  
Automotive Manufacturing ◽  
Customer Preferences ◽  
Deterministic Function ◽  
State Of Mind ◽  
Utility Distribution ◽  
Decision Based Design

Abstract Engineering and manufacturing abilities of firms evolve with every passing year and so do the preferences of the customers buying their products. Reconciling this coevolution is essential to staying competitive in the marketplace. In this paper, we provide a looped Bayesian framework to accomplish this so that designs can evolve as engineering capabilities increase and customer preferences change. We begin with an approach to incorporating the voice of the customer through the multi-attribute utility function, the core of decision-based design. We consider the utility to be a stochastic function governed by shape parameters that are random variables. Typically, a representative preference or utility function is used or the function is aggregated over many decision makers and regarded as a deterministic function of specified shape parameters. In our approach, the shape parameters represent the stochastic nature of preference behavior either due to variation in a decision maker’s state of mind from one decision to another, or due to a multiplicity of decision makers. The novelty of this approach is in taking a Bayesian perspective on the stochastic utility function. We consider the utility distribution in the design phase as a prior distribution and we update the prior to a posterior with feedback on the actual product in production. The method is valuable in providing a means to improve the level of informativeness of the design level utility function for adjustments to the design or for the next design revision in the cycle of continuous improvement. We present our approach on a real-life assembly problem in an automotive manufacturing floor.

A Multiattribute Utility Approach to Target Assignment

2009 ◽  
Author(s):  
Xi Chen ◽  
Ali E. Abbas ◽  
Dusˇan M. Stipanovic´
Keyword(s):  
Utility Function ◽  
Kinematic Model ◽  
Multiattribute Utility ◽  
Control Lyapunov Function ◽  
Control Laws ◽  
Function Design ◽  
Target Assignment ◽  
Conduct Sensitivity Analysis ◽  
Multi Agent ◽  
Utility Approach

This paper introduces the multiattribute utility theory to the control Lyapunov function design framework. As an illustration we focus on the problem of multi-target assignment. With this formulation, we use a global multiattribute utility function as a multivariate objective function that should be minimized for the agents to achieve their objectives. The objectives represent deviations of each agent from specified targets. We provide closed form feedback control laws, based on the multiattribute utility function, for general nonlinear multi-agent system models affine in control. Finally, we present simulation results and conduct sensitivity analysis for two different models that are affine in control; basic kinematic model and nonlinear nonholonomic unicycle model.

Sign in / Sign up

Export Citation Format

Share Document