Abstract | ||
---|---|---|
Schema learning is a way to discover probabilistic, constructivist, pre- dictive action models (schemas) from experience. It includes meth- ods for finding and using hidden state to make predictions mor e accu- rate. We extend the original schema mechanism (1) to handle arbitrary discrete-valued sensors, improve the original learning cr iteria to handle POMDP domains, and better maintain hidden state by using schema pre- dictions. These extensions show large improvement over the original schema mechanism in several rewardless POMDPs, and achievevery low prediction error in a difficult speech modeling task. Furthe r, we compare the extended schema learner to the recently introduced predictive state representations (2), and find their predictions of next-ste p action effects to be approximately equal in accuracy. This work lays the foundation for a schema-based system of integrated learning and planning. |
Year | Venue | Keywords |
---|---|---|
2004 | NIPS | prediction error |
Field | DocType | Citations |
Constructivism (philosophy of education),Integrated learning,Mean squared prediction error,Partially observable Markov decision process,Computer science,Speech modeling,Learning experience,Artificial intelligence,Probabilistic logic,Schema (psychology),Machine learning | Conference | 16 |
PageRank | References | Authors |
0.94 | 9 | 2 |
Name | Order | Citations | PageRank |
---|---|---|---|
Michael P. Holmes | 1 | 115 | 7.15 |
Charles L. Isbell | 2 | 504 | 65.79 |