Top-K Quality Plan Generation via Iterative Forbidding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current top-k planning methods focus on generating a set of plans with a specified number of high quality, but struggle with semantic equivalence and the computational burden of generating all valid orderings, limiting their effectiveness in finding optimal sets of plans with a quality bound.
Innovation Solution
The approach introduces a quality bound for plan validity, allowing for the representation of sets of valid plan re-orderings with a single plan, using iterative operations to reformulate the planning problem and forbid identified plans and their reorderings, thereby focusing on high-quality plans with the lowest cost and implicitly representing equivalence classes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If traditional top-k planning methods generate a specified number of high quality plans, then the quantity of plans is improved, but the computational burden and complexity increase significantly
Solution Approach 1:
The patent changes the parameter from generating a fixed number of plans (k) to generating plans within a quality bound (cost threshold). This shifts the optimization criterion from quantity-based to quality-based, reducing computational complexity by avoiding enumeration of all valid orderings while ensuring plans meet the cost bound q.
Solution Approach 2:
Instead of generating k plans and then filtering by quality, the patent inverts the approach by first establishing a quality bound q and generating plans that satisfy this bound. This inversion allows the system to focus computational resources on finding high-quality plans rather than generating and filtering a fixed number of plans.
2Reliability
If all valid plan orderings are generated to ensure completeness, then the reliability is improved, but the loss of time and computational resources increases
Solution Approach 1:
The patent performs preliminary action by establishing the quality bound q before plan generation. By pre-defining the cost threshold and using it to guide the search process, the system avoids generating plans that would exceed the bound, thereby reducing computation time while maintaining reliability through the quality guarantee.
Solution Approach 2:
The patent extracts and forbids identified plans and their equivalent reorderings from further consideration in iterative operations. This extraction approach ensures completeness by systematically eliminating already-found solutions while avoiding redundant computation of semantically equivalent plans, thus reducing time loss.
3Manufacturing precision
If semantic equivalence of plans is considered to avoid redundancy, then the manufacturing precision is improved, but the device complexity increases
Solution Approach 1:
The patent uses an intermediary approach by introducing a planning graph representation and iterative forbidding mechanism. Instead of directly comparing semantic equivalence of all plan pairs (which would be complex), the system uses the planning graph and quality bound as intermediaries to implicitly handle equivalence, reducing algorithmic complexity while maintaining precision.
4Manufacturing precision
If the focus is on generating plans with the lowest cost, then the manufacturing precision is improved, but the quantity of plans generated decreases
Solution Approach 1:
The patent introduces dynamics by iteratively adjusting the set of forbidden plans based on identified high-quality plans. The system dynamically reformulates the planning problem in subsequent iterations, adapting the search space to focus on remaining high-quality plans while maintaining the quality bound, thus balancing precision and quantity.
Data Source
AI summary
Embodiments are provided for providing top-K quality plans streaming applications in a computing environment. A set of top-K quality plans using a quality bound for a planning problem. The planning problem may be reformulated in one or more subsequent iterations and forbidding use one or more of the set of top-K quality plans. Identifying one or more of the set top-K quality plans having a quality less than the quality bound during the one or more subsequent iterations.