Add abstract
Want to add your dissertation abstract to this database? It only takes a minute!
Search abstract
Search for abstracts by subject, author or institution
Want to add your dissertation abstract to this database? It only takes a minute!
Search for abstracts by subject, author or institution
Integrating logical reasoning and probabilistic graphical models for spokendialog system
by Han Xu
Institution: | Texas Tech University |
---|---|
Year: | 2017 |
Keywords: | SDS; POMDP; ASP; P-log |
Posted: | 02/01/2018 |
Record ID: | 2152130 |
Full text PDF: | http://hdl.handle.net/2346/72616 |
In recent times, partially observable Markov decision processes (POMDPs), aninstance of probabilistic sequential decision making under partialobservability, have been used to create SDS. These systems use the inherentcapabilities of POMDPs to model (and account for) the common causes of speechrecognition errors such as background noise and variations in user input.However, POMDP-based solutions can become computationally intractable as thenumber of possible states becomes large. This is typically the case for dialogin many practical application domains. These systems also fail to exploit thecontextual commonsense knowledge about the domain or the sequence of possibleutterances in dialog in any specific domain.This thesis is a significant step towards addressing the challenges indesigning SDS for practical applications. The proposed novel architectureimproves the tractability of SDSs by exploiting contextual domain knowledge,and the complementary strengths of commonsense inference and probabilisticsequential decision making. Specifically, we use Plog, a probabilisticextension to the declarative language Answer Set Prolog (ASP), to represent andreason with domain knowledge. We also use a POMDP to select the actionsexecuted by the SDS, i.e., the verbal utterances of the system and theconclusions drawn, in response to the verbal inputs of the user. Commonsenseinference with the domain knowledge is used to identify the subset of the statespace that needs to be considered by the POMDP, and to revise the probabilisticbeliefs currently encoded by the POMDP. The updated probabilistic beliefdistributions are, in turn, used to revise the domain knowledge for subsequentinference. The architecture's capabilities are illustrated and evaluated in thecontext of a system that accepts verbal instructions from a human to move objects between specific locations.Advisors/Committee Members: Sridharan, Mohan (committee member), Mengel, Susan (committee member), Watson, Richard (Committee Chair).
Want to add your dissertation abstract to this database? It only takes a minute!
Search for abstracts by subject, author or institution
Electric Cooperative Managers' Strategies to Enhan...
|
|
Bullied!
Coping with Workplace Bullying
|
|
The Filipina-South Floridian International Interne...
Agency, Culture, and Paradox
|
|
Solution or Stalemate?
Peace Process in Turkey, 2009-2013
|
|
Performance, Managerial Skill, and Factor Exposure...
|
|
The Deritualization of Death
Toward a Practical Theology of Caregiving for the ...
|
|
Emotional Intelligence and Leadership Styles
Exploring the Relationship between Emotional Intel...
|
|
Commodification of Sexual Labor
Contribution of Internet Communities to Prostituti...
|
|
The Census of Warm Debris Disks in the Solar Neigh...
|
|
Risk Factors and Business Models
Understanding the Five Forces of Entrepreneurial R...
|
|