自主性代理人協商模型與學習機制之發展  -以半導體測試排程為例

在動態賽局中使用自動化談判是一極具效益的研究議題，可運用於協同式供應鏈管理與電子商務等領域。本研究探討一對一自動化談判模型，在服務者端，建立自簡單到複雜的五種不同型態代理人，分別為：「隨機型」、「合理型」、「利他利己性合作型」、「利他性合作型」與「學習型」代理人。顧客端為隨機型代理人，於每次協商中產生一組可行排程，再提出協商值；合理型代理人以「協商決策函式(Negotiation Decision Function, NDF)」產生談判值後，再以基因演算法產生一組可行排程協商值，且此協商值之得分不低於NDF所產生之談判值，並提出此提議；利己利他性合作型代理人以NDF模組配合「協商權衡機制」，以NDF提出談判值，再以基因演算法產生一組可行排程後，此排程協商值會透過協商權衡機制，以雙方提議之得分值差距做有限度的讓步；利他性合作型代理人以NDF產生談判值後，以基因演算法產生一組以對方角度尋求最佳得分之可行排程；學習型代理人透過感知機類神經網路，藉由協商過程所獲得的資訊，預測協商對手的偏好與議題權重。本研究以一個在靜態環境下的半導體測試排程為例，經由協商過程，學習到對方的議題權重，達到加速溝通協商的目的。

關鍵字

自主性代理人；協商機制；生產排程；機器學習

並列摘要

Automated negotiation is an active research domain in dynamic games. This study focuses on an one-to-one automated negotiation model, and develops five different server agents. The five different agents are Random Agent, Rational Agent, Cooperative Agent with trade-offs, Cooperative Agent and Learning Agent. A random agent will propose a feasible schedule to its client. A rational agent generates a negotiation proposal using negotiation decision function (NDF). And generates a proposal by a generic algorithm (GA) in which its score is not lower than the score of negotiation proposal. A cooperative agent with trade-offs generates a negotiation proposal by NDF and the trade-offs mechanism. A naive cooperative agent also generates a negotiation proposal by NDF, and generates an available scheduling by GA which seeking the maximize utility of the other agent. Finally, a learning agent will predict the weights of negotiation issues of its opponent by perceptron neural network. This study applied the proposal method to the semiconductor testing scheduling and its experimental outcomes showed a promising results.

並列關鍵字

Scheduling ； Machine Learning ； Negotiation Mechanism ； Autonomous Agent

參考文獻

Oprea, M., (2002), “An Adaptive Negotiation Model for Agent-Based Electronic Commerce”, Studies in Informatics and Control, 11(3), 271-279.

古漢舜 (2002)，「多資源限制下智慧型代理人協商機制之設計—以半導體廠作業為例」，碩士論文，私立中原大學

Bond, A. H., and Gasser, L., eds. (1988), “A Multi-Agent Incremental Negotiation Scheme for Meetings Scheduling.” In proceedings ANZIIS-95, the Third Australian and New Zealand Conference on Intelligent Information Systems, __.

Bui, H. H., D. Kieronska and Venkatesh, S. (1996), “Learning Other Agents’ Preferences in Multi-agent Negotiation”, In Proceedings of the National Conference on Artificial Intelligence(AAAI-96), 114-119.

Chiu, C. and Yih, Y.,(1995) “A learning-based methodology for dynamic scheduling in distributed manufacturing systems.” International Journal of Production Research, 33 (11), 3217-3232.

國際替代計量

自主性代理人協商模型與學習機制之發展 -以半導體測試排程為例

未授權

主題瀏覽