Multimedia communication in mobile and ad hoc networks used by real time applications can be improved by adding intelligent and adaptive cababilities. This new functionality will allow them to adapt to contantly and unpredictably changing network conditions. Derived from this adaptivity, the user will perceive a more or less constant quality instead of the high variable quality perceived in nowadays applications. In this work, we maintain the following thesis: both machine learning and intelligent agents will play an important role in the improvement of the aplications we mentioned above. Machine learning, by means of reinforcement learning will provide adaptivity. Intelligent agents will ease P2P computation. This paper focuses on approaches for both topics.