The Emergence of Individuality
This work addresses multi-agent cooperation by promoting individuality, which could enhance efficiency in domains like robotics or gaming, though it appears incremental as it builds on existing MARL algorithms.
The paper tackles the problem of fostering individuality in multi-agent reinforcement learning to improve cooperation, proposing a method that uses a probabilistic classifier with intrinsic rewards and regularizers, and reports significant performance gains over existing methods in various cooperative scenarios.
Individuality is essential in human society, which induces the division of labor and thus improves the efficiency and productivity. Similarly, it should also be the key to multi-agent cooperation. Inspired by that individuality is of being an individual separate from others, we propose a simple yet efficient method for the emergence of individuality (EOI) in multi-agent reinforcement learning (MARL). EOI learns a probabilistic classifier that predicts a probability distribution over agents given their observation and gives each agent an intrinsic reward of being correctly predicted by the classifier. The intrinsic reward encourages the agents to visit their own familiar observations, and learning the classifier by such observations makes the intrinsic reward signals stronger and the agents more identifiable. To further enhance the intrinsic reward and promote the emergence of individuality, two regularizers are proposed to increase the discriminability of the classifier. We implement EOI on top of popular MARL algorithms. Empirically, we show that EOI significantly outperforms existing methods in a variety of multi-agent cooperative scenarios.