Wizard: Suggesting Macro-Actions Comprehensively
نویسندگان
چکیده
This paper presents Wizard, a generalised framework for learning macro-actions in planning. Wizard suggests macroactions that can be provided as additional actions for future planning. It enhances a domain for a planner through comprehensive macro suggestions. Wizard learns macro-actions for arbitrary planners or domains without exploiting their structural properties. It not only captures macro-actions that are observable from examples, but also evolves other macroactions that are not observable. It learns macro-actions that capture various system aspects. It explores both individual macro-actions and their combinations.
منابع مشابه
Simulating the use of Macro-Actions Through Action Reordering
The use of macro-actions in planning introduces a trade-off. Macro-actions can offer search guidance by suggesting sequences of actions; but can potentially make search more expensive by increasing the branching factor. In this paper we present a technique for simulating the use of macro-actions by altering the order in which actions are considered for application during enforced hill-climbing ...
متن کاملOnline generation and use of macro-actions in forward-chaining planning
This thesis presents a technique for online learning and management of macroactions in forward chaining planning. Macro-actions are learnt on plateaux, areas of the search landscape where the heuristic cannot offer good search guidance, and are reused when future plateaux are encountered. Libraries of macro-actions are generated, storing macro-actions for use on future problems. Several strateg...
متن کاملSeeing What You Said: How Wizards Use Voice Search Results
A Wizard-of-Oz experiment investigates how voice search could affect dialogue management strategies. The study design has two novel components. First, a single turn exchange is examined, rather than a full dialogue. Second, wizards partner with a dialogue system, so internal system features unavailable to the wizard can be used to model wizard actions. Wizards see the output of automated speech...
متن کاملPUMA: Planning Under Uncertainty with Macro-Actions
Planning in large, partially observable domains is challenging, especially when a long-horizon lookahead is necessary to obtain a good policy. Traditional POMDP planners that plan a different potential action for each future observation can be prohibitively expensive when planning many steps ahead. An efficient solution for planning far into the future in fully observable domains is to use temp...
متن کاملOnline Identification of Useful Macro-Actions for Planning
This paper explores issues encountered when performing online management of large collections of macro-actions generated for use in planning. Existing approaches to managing collections of macro-actions are designed for use with offline macro-action learning, pruning candidate macro-actions on the basis of their effect on the performance of the planner on small training problems. In this paper ...
متن کامل