论文信息 - Transfer Learning for Bilateral Multi Issue Negotiation

Transfer Learning for Bilateral Multi Issue Negotiation

This paper proposes a novel strategy named Transfer between Negotiation Tasks (TNT) for automated bilateral negotiation with multiple issues. TNT is able to probabilistically transfer between different negotiation tasks in order to bias the target agent’s learning behavior towards improved performance without unrealistic assumptions. We analyze the performance of our strategy and show that it substantially outperforms a powerful negotiation strategy across a variety of negotiation scenarios.

[1] Sarit Kraus,et al. Genius: negotiation environment for heterogeneous agents , 2009, AAMAS.

[2] Carl E. Rasmussen,et al. Gaussian processes for machine learning , 2005, Adaptive computation and machine learning.

[3] Jacques L. Koko,et al. The Art and Science of Negotiation , 2009 .

[4] Nicholas R. Jennings,et al. Learning on opponent's preferences to make effective multi-issue negotiation trade-offs , 2004, ICEC '04.

[5] Gregory E. Kersten,et al. Predicting opponent's moves in electronic negotiations using neural networks , 2008, Expert Syst. Appl..

[6] Ryszard Kowalczyk,et al. Predicting partner's behaviour in agent negotiation , 2006, AAMAS '06.

[7] Haitham Bou-Ammar,et al. Reinforcement learning transfer via sparse coding , 2012, AAMAS.

[8] A. Rubinstein. Perfect Equilibrium in a Bargaining Model , 1982 .

[9] Qiang Yang,et al. A Survey on Transfer Learning , 2010, IEEE Transactions on Knowledge and Data Engineering.

[10] Gerhard Weiss,et al. An Efficient and Adaptive Approach to Negotiation in Complex Environments , 2012, ECAI.

[11] Sandip Sen,et al. Modeling opponent decision in repeated one-shot negotiations , 2005, AAMAS '05.