ANEGMA: an automated negotiation model for e-markets

Abstract We present a novel negotiation model that allows an agent to learn how negotiate during concurrent bilateral negotiations in unknown and dynamic e-markets. The uses actor-critic architecture with model-free reinforcement learning strategy expressed as deep neural network. pre-train the by supervision from synthetic market data, thereby decreasing exploration time required for negotiation. As result, we can build automated agents adapt different e-market settings without need be pre-programmed. Our experimental evaluation shows our based outperform two existing well-known strategies one-to-many range of settings.