I'm looking for literature on the application of Reinforcement Learning (RL) algorithms in a business context. Most articles and examples in books on RL are about the application of RL to games (Backgammon, Chess, Atari games, Go) and/or toy problems (Cliff walking, Windy Grid, Mountain car, etc). Of course these are convenient contexts, because playing a game can relatively easily be simulated on a computer, thereby producing "unlimited" amounts of transitions for learning and testing.
In business environments there generally is no such luxury - unless the process in question is simulated. Also, it is the question whether results obtained in playing games can be 'tranferred' to business contexts without further research (I think not).
So, I'm looking for references to RL used for optimizing business processes, such as processes involving multiple sequential communications (possibly via different channels) with clients, as can be found in marketing, acquisition, debt collection, or the like.
Until now I've found some articles by IBM/Naoki Abe et al., e.g.:
- ‘Sequential Cost Sensitive Decision Making with Reinforcement Learning’,
- ‘Optimizing debt collections using constrained reinforcement learning’
(https://researcher.watson.ibm.com/researcher/view_person_subpage.php?id=6746), but I was wondering if anyone can point me to additional and/or more recent literature.