TY - CONF
T1 - Combining reinforcement learning with information-state update rules
AU - Heeman, Peter A.
N1 - Funding Information:
With regard to financial support, in the state of Victoria, the State Government of Victoria provides funding in the form of Victorian Landcare Grants. At the national level, the federal government provides funding through National Landcare Program Regional Funding and also provides the possibility of tax deductions for landowners incurring capital expenditures for their property improvements (known as a Landcare tax benefit).
Funding Information:
According to the documentation distributed, the event was a project launch event. The project, entitled “Demonstrating Sustainable Farm Practices in Western Port, Port Phillip and Yarra catchment” is “managed by the Western Port Catchment Landcare Network and supported by Port Phillip & Western CMA through funding from the Australian Government”. The “aim of this project is to work with farmers from all enterprises to promote and support the uptake of sustainable farming practices. Landcare staff will primarily deliver the project but we will utilise the services of farm consultants, agronomists, DEPI and soil scientists when required.”
PY - 2007
Y1 - 2007
N2 - Reinforcement learning gives a way to learn under what circumstances to perform which actions. However, this approach lacks a formal framework for specifying hand-crafted restrictions, for specifying the effects of the system actions, or for specifying the user simulation. The information state approach, in contrast, allows system and user behavior to be specified as update rules, with preconditions and effects. This approach can be used to specify complex dialogue behavior in a systematic way. We propose combining these two approaches, thus allowing a formal specification of the dialogue behavior, and allowing hand-crafted preconditions, with remaining ones determined via reinforcement learning so as to minimize dialogue cost.
AB - Reinforcement learning gives a way to learn under what circumstances to perform which actions. However, this approach lacks a formal framework for specifying hand-crafted restrictions, for specifying the effects of the system actions, or for specifying the user simulation. The information state approach, in contrast, allows system and user behavior to be specified as update rules, with preconditions and effects. This approach can be used to specify complex dialogue behavior in a systematic way. We propose combining these two approaches, thus allowing a formal specification of the dialogue behavior, and allowing hand-crafted preconditions, with remaining ones determined via reinforcement learning so as to minimize dialogue cost.
UR - http://www.scopus.com/inward/record.url?scp=70349253156&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=70349253156&partnerID=8YFLogxK
M3 - Paper
AN - SCOPUS:70349253156
SP - 268
EP - 275
T2 - Human Language Technologies 2007: The Conference of the North American Chapter of the Association for Computational Linguistics, NAACL HLT 2007
Y2 - 22 April 2007 through 27 April 2007
ER -