Markov decisions processes (MDPs) and their model-free counterpart in reinforcement learning (RL) have known a large success in the last two decades. Although research in these two areas has been taking place for more than fifty years, the field gained momentum only recently following the advent of powerful hardware and algorithms with which supra-human performance were obtained in games like Chess or Go. However, these impressive successes often rely on quite exceptional hardware possibilities and cannot be applied in many “usual” contexts where, for instance, the volume of data available or the amount of computing power is more restricted. To define the next generation of more “democratic” and widely applicable algorithms, such methods still need to deal with very demanding exploration issues as soon as the state/action spaces are not small. One way around this is to use underlying knowledge and structure present in many MDPs. This is especially true for problems related to scheduling and resource sharing in systems like server farms, clouds, and cellular wireless networks. In recent years, there has been a huge research effort improving the efficiency of learning algorithms by leveraging the structure of the underlying problem, both in the model-based and model-free frameworks. This workshop will gather experts and students working at the frontier of this topic.
Researchers who wish to present their work at the workshop RL4SN should submit an abstract (for more details see Instructions for Authors). Participants at the conference will be given the opportunity to submit a paper based on their talk to a special issue of the journal Queueing Systems: Theory and Applications (QUESTA) (other journals are pending approval). For more details, see the Special Issues page.
This workshop is part of the thematic semester Stochastic control and learning for complex networks (SOLACE).