 |
ML Newsnew | past | best | rss | submit | about |
| Reinforcement Learning is not a Causal problem (arxiv.org) | |
| 1 point by sel1 on Aug 22, 2019 | hide | past | pdf | discuss on HN | |
|
In plain words: It compares two kinds of structure on the same set of facts—what goes with what versus what causes what—to show they are not the same. By that logic, today's reinforcement learning is not a causal problem, even though an agent chooses actions. AbstractWe use an analogy between non-isomorphic mathematical structures defined over the same set and the algebras induced by associative and causal levels of information in order to argue that Reinforcement Learning, in its current formulation, is not a causal problem, independently if the motivation behind it has to do with an agent taking actions. | |
| add comment on HN |