PPT Slide
Adaptive Agents for Negotiation
- In our first scenaria agents negotiate 1 issue : price
1. MTA (time dependent b =1) Negotiating with ABA
2. ABA competing with 1 MTA; negotiating with buyer MTA
Action selection - exploitation vs. exploration
? -greedy: a small probability ? of uniformly choosing a non-greedy action
Softmax: a degree of exploration ? for choosing actions according to their ranking
Choosing the best action (Softmax approach):
Where: p(a) - probability of choosing action a