In complex, dynamic environments, an agent's knowledge of the environment (its domain knowled... more In complex, dynamic environments, an agent's knowledge of the environment (its domain knowledge) will rarely be complete and correct. Existing approaches to learning and correcting domain knowledge have focused on either learning procedural knowledge to directly guide execution (e.g. reinforcement learners) or learning declarative planning knowledge (e.g. theory revision systems). Systems that only learn execution knowledge are generally only applicable to small domains. In these domains it is possible to learn an execution policy that covers the entire state space, making planning unnecessary. Conversely, existing approaches to learning declarative planning knowledge are applicable to large domains, but they are limited to simple agents, where actions produce immediate, deterministic effects in fully sensed, noise-free environments, and where there are no exogenous events. This research investigates the use of procedural knowledge to support the learning of planning knowledge i...
We describe a learning from diagrammatic behavior specifications approach, where the task-perform... more We describe a learning from diagrammatic behavior specifications approach, where the task-performance knowledge of a human expert is transferred to an agent program using abstract behavior scenarios that the expert and the agent program interactively specify. The diagrammatic interface serves as a communication medium between the expert and the agent program to share knowledge during behavior specification. A relational learning by observation component interprets these scenarios in the context of background knowledge and expert annotations to learn first-order rules that represent the task-performance knowledge for an improved agent program.
ABSTRACT: The last ten years has seen a revolution in the complexity and realism of human ,behavi... more ABSTRACT: The last ten years has seen a revolution in the complexity and realism of human ,behavior models (HBMs). However, the cost of developing realistic HBMs continues to increase as much of the detailed and complex knowledge must be manually encoded to produce realisticbehavior. The focus of this project is on reducing the cost of acquiring, validating and maintaining the knowledge used in realistic HBMs. Our approach is to develop tools that allow subject matter experts (SMEs) to specify behavior using abstract scenarios represented as diagrams. The SME can graphically describe the conditions under which actions and goals should be pursued, together with the associated reasons for those decisions. The system, guided by the expert’s choices, analyzes and automatically generalizes from the example scenarios, alerting the SME to inconsistencies and missing knowledge. The system incrementally generates an executable HBM whose behavior the SME can view and modify during development...
... exciting goal for AI, a general intelligent agent. It's great to work with smart people ... more ... exciting goal for AI, a general intelligent agent. It's great to work with smart people and, of course, John's a very smart guy, but there are a lot of smart people in AI or any academic eld. John's di erent because he combines a fun ...
The last ten years has seen a revolution in the complexity and realism of human behavior models (... more The last ten years has seen a revolution in the complexity and realism of human behavior models (HBMs). However, the cost of developing realistic HBMs continues to increase as much of the detailed and com- plex knowledge must be manually encoded to produce realistic behavior. The focus of this project is on reducing the cost of acquiring, validating and maintaining
In complex, dynamic environments, an agent's domain knowledge will rarely becomplete and corr... more In complex, dynamic environments, an agent's domain knowledge will rarely becomplete and correct. Existing deliberate approaches to domain theory correction aresignificantly restricted in the environments where they can be used. These systems aretypically not used in agent-based tasks and rely on declarative representations to supportnon-incremental learning. This research investigates the use of procedural knowledge tosupport deliberate incremental error correction in
We describe a learning from diagrammatic behavior specifications approach, where the task-perform... more We describe a learning from diagrammatic behavior specifications approach, where the task-performance knowledge of a human expert is transferred to an agent program using abstract behavior scenarios that the expert and the agent program interactively specify. The diagrammatic interface serves as a communication medium between the expert and the agent program to share knowledge during behavior specification. A relational learning
In complex, dynamic environments, an agent's knowledge of the environment (its domain knowled... more In complex, dynamic environments, an agent's knowledge of the environment (its domain knowledge) will rarely be complete and correct. Existing approaches to learning and correcting domain knowledge have focused on either learning procedural knowledge to directly guide execution (e.g. reinforcement learners) or learning declarative planning knowledge (e.g. theory revision systems). Systems that only learn execution knowledge are generally only applicable to small domains. In these domains it is possible to learn an execution policy that covers the entire state space, making planning unnecessary. Conversely, existing approaches to learning declarative planning knowledge are applicable to large domains, but they are limited to simple agents, where actions produce immediate, deterministic effects in fully sensed, noise-free environments, and where there are no exogenous events. This research investigates the use of procedural knowledge to support the learning of planning knowledge i...
We describe a learning from diagrammatic behavior specifications approach, where the task-perform... more We describe a learning from diagrammatic behavior specifications approach, where the task-performance knowledge of a human expert is transferred to an agent program using abstract behavior scenarios that the expert and the agent program interactively specify. The diagrammatic interface serves as a communication medium between the expert and the agent program to share knowledge during behavior specification. A relational learning by observation component interprets these scenarios in the context of background knowledge and expert annotations to learn first-order rules that represent the task-performance knowledge for an improved agent program.
ABSTRACT: The last ten years has seen a revolution in the complexity and realism of human ,behavi... more ABSTRACT: The last ten years has seen a revolution in the complexity and realism of human ,behavior models (HBMs). However, the cost of developing realistic HBMs continues to increase as much of the detailed and complex knowledge must be manually encoded to produce realisticbehavior. The focus of this project is on reducing the cost of acquiring, validating and maintaining the knowledge used in realistic HBMs. Our approach is to develop tools that allow subject matter experts (SMEs) to specify behavior using abstract scenarios represented as diagrams. The SME can graphically describe the conditions under which actions and goals should be pursued, together with the associated reasons for those decisions. The system, guided by the expert’s choices, analyzes and automatically generalizes from the example scenarios, alerting the SME to inconsistencies and missing knowledge. The system incrementally generates an executable HBM whose behavior the SME can view and modify during development...
... exciting goal for AI, a general intelligent agent. It's great to work with smart people ... more ... exciting goal for AI, a general intelligent agent. It's great to work with smart people and, of course, John's a very smart guy, but there are a lot of smart people in AI or any academic eld. John's di erent because he combines a fun ...
The last ten years has seen a revolution in the complexity and realism of human behavior models (... more The last ten years has seen a revolution in the complexity and realism of human behavior models (HBMs). However, the cost of developing realistic HBMs continues to increase as much of the detailed and com- plex knowledge must be manually encoded to produce realistic behavior. The focus of this project is on reducing the cost of acquiring, validating and maintaining
In complex, dynamic environments, an agent's domain knowledge will rarely becomplete and corr... more In complex, dynamic environments, an agent's domain knowledge will rarely becomplete and correct. Existing deliberate approaches to domain theory correction aresignificantly restricted in the environments where they can be used. These systems aretypically not used in agent-based tasks and rely on declarative representations to supportnon-incremental learning. This research investigates the use of procedural knowledge tosupport deliberate incremental error correction in
We describe a learning from diagrammatic behavior specifications approach, where the task-perform... more We describe a learning from diagrammatic behavior specifications approach, where the task-performance knowledge of a human expert is transferred to an agent program using abstract behavior scenarios that the expert and the agent program interactively specify. The diagrammatic interface serves as a communication medium between the expert and the agent program to share knowledge during behavior specification. A relational learning
Uploads
Papers by Douglas Pearson