Negotiating Agent Adaptation via Utility Function Change Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing automated negotiation systems become less competent when the utility function of an opposing agent changes during frequent negotiations, as they are not adapted to handle such changes effectively.
Innovation Solution
A parameter-based transfer learning approach is employed, where a negotiating agent is trained to detect changes in the opposing agent's utility function and adapt by generating training samples and updating its negotiation strategy model, enabling it to improve over time and negotiate more efficiently.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a traditional automated negotiation system uses a fixed negotiation strategy model, then the system operates with stable and predictable behavior, but the system becomes less competent when the utility function of the opposing agent changes
Solution Approach 1:
The negotiation strategy model is transformed from a fixed static structure to a dynamic one that can adapt its parameters based on detected changes in the opposing agent's utility function. The system continuously monitors negotiation outcomes and adjusts its strategy model parameters accordingly, enabling it to maintain competence despite changes in the opposing agent's preferences or objectives.
Solution Approach 2:
The system changes the parameters of its negotiation strategy model based on detected utility function changes in the opposing agent. By monitoring negotiation patterns and outcomes, the system identifies when the opposing agent's utility function has changed and adjusts its own strategy parameters (such as concession rates, offer timing, and issue prioritization) to maintain effective negotiation performance.
2Productivity
If the negotiating agent continuously learns and adapts to changes in utility function, then negotiation efficiency and benefit are maintained, but the complexity of the negotiation system increases
Solution Approach 1:
The system implements a feedback mechanism where negotiation outcomes are continuously monitored and fed back to the strategy model. This feedback loop enables the system to detect changes in the opposing agent's utility function and adjust its strategy accordingly. The feedback-based approach maintains negotiation efficiency without requiring complete system redesign, as it builds adaptation capabilities into the existing negotiation framework.
Solution Approach 2:
The system performs preliminary detection of utility function changes by monitoring early negotiation patterns and outcomes. By detecting changes early in the negotiation process, the system can proactively adjust its strategy before significant performance degradation occurs, maintaining efficiency without requiring complex real-time reconfiguration mechanisms.
3Adaptability or versatility
If the negotiation system detects changes in utility function and generates training samples, then the agent adapts to new negotiation conditions, but the time and computational resources required increase
Solution Approach 1:
The system applies partial adaptation by generating training samples and updating the strategy model only when changes in the opposing agent's utility function are detected above a certain threshold. This selective adaptation approach avoids the time and computational costs of continuous full retraining while maintaining adaptability to significant changes in negotiation conditions.
Solution Approach 2:
The system performs utility function change detection and strategy model updates at periodic intervals based on accumulated negotiation data rather than continuously. This periodic action reduces the time and computational resources required for adaptation while still maintaining current effectiveness in changing negotiation environments.
Data Source
AI summary
Automated negotiation agent adaptation is performed by detecting change in a utility function involved in automated negotiation between a supporting agent and an opposing agent while the supporting agent operates according to a first negotiation strategy model, generating a plurality of training samples from automated negotiation between the supporting agent and the opposing agent while the supporting agent operates according to a baseline negotiation strategy model, and training an initialized negotiation strategy model using the plurality of training samples to produce a second negotiation strategy model.