The Compressed Differential Heuristic

Goldenberg, Meir; Sturtevant, Nathan R.; Felner, Ariel; Schaeffer, Jonathan

doi:10.1609/aaai.v25i1.7823

Cited by 5 publications

(5 citation statements)

References 4 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…COLT is also a versatile data structure with potential application to other search and planning problems, e.g., utilizing the hierarchical search to find optimal meeting points for ride-sharing. Further improvement of search heuristics may also be possible, e.g., in the direction of Compressed Differential Heuristics (Goldenberg et al 2011).…”

Section: Discussionmentioning

confidence: 99%

“…Landmark Lower-Bounds (LLBs): Lower-bounds based on landmarks, also called differential heuristics (Goldenberg et al 2011), involve selecting a set L of m "landmark" vertices and then pre-computing distances from each landmark to all vertices in V . Given two vertices q and p, a lowerbound on network distance can be computed by (1) using the distances to landmark l i and the triangle inequality.…”

Section: Preliminariesmentioning

confidence: 99%

“…One of the most popular heuristics to estimate network distances is landmark-based lower-bounds (LLBs) (Goldberg and Harrelson 2005) also often referred to as differential heuristics (Sturtevant et al 2009;Goldenberg et al 2011). (Kriegel et al 2008) proposes a hierarchical landmark scheme to reduce index space cost.…”

Section: Related Workmentioning

confidence: 99%

See 2 more Smart Citations

Hierarchical Graph Traversal for Aggregate k Nearest Neighbors Search in Road Networks

Abeywickrama

Cheema

Storandt

2020

ICAPS

View full text Add to dashboard Cite

Location-based services rely heavily on efficient methods that search for relevant points-of-interest (POIs) close to a given location. A k nearest neighbors (kNN) query is one such example that finds k closest POIs from an agent's location. While most existing techniques focus on finding nearby POIs for a single agent, many applications require POIs that are close to multiple agents. In this paper, we study a natural extension of the kNN query for multiple agents, namely, the Aggregate k Nearest Neighbors (AkNN) query. An AkNN query retrieves k POIs with the smallest aggregate distances where the aggregate distance of a POI is obtained by aggregating its distances from the multiple agents (e.g., sum of its distances from each agent). Existing search heuristics are designed for a single agent and do not work well for multiple agents. We propose a novel data structure COLT (Compacted Object-Landmark Tree) to address this gap by enabling efficient hierarchical graph traversal. We then utilize COLT for a wide range of aggregate functions to efficiently answer AkNN queries. In our experiments on real-world and synthetic data sets, our techniques significantly improve query performance, typically outperforming existing approaches by more than an order of magnitude in almost all settings.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Preliminariesmentioning

confidence: 99%

See 1 more Smart Citation

Hierarchical Graph Traversal for Aggregate k Nearest Neighbors Search in Road Networks

Abeywickrama

Cheema

Storandt

2020

ICAPS

View full text Add to dashboard Cite

show abstract

“…In addition, preprocessing is also common in the heuristic search community. For example, it is used in the construction of pattern databases heuristics (Culberson and Schaeffer 1998;Korf 1997;Felner, Korf, and Hanan 2004;Holte et al 2006;Haslum et al 2007), heuristics for 2D-pathfinding problems (Sturtevant et al 2009;Pochter et al 2010;Goldenberg et al 2011), learning-based heuristics (Ernandes and Gori 2004;Samadi, Felner, and Schaeffer 2008; Jabbari Arfaee, Zilles, and Holte 2011), search effort prediction formulas (Korf, Reid, and Edelkamp 2001;Zahavi et al 2010;Lelis, Zilles, and Holte 2011) and search cost prediction formulas (Lelis, Stern, and Jabbari Arfaee 2011;Lelis et al 2012).…”

Section: P R( Hmentioning

confidence: 99%

Search-Aware Conditions for Probably Approximately Correct Heuristic Search

Stern

Felner

Holte

2021

SOCS

View full text Add to dashboard Cite

The notion of finding a solution that is approximately optimal with high probability was recently introduced to the field of heuristic search,formalized as Probably Approximately Correct Heuristic Search, or PAC search in short. A big challenge when constructing a PAC search algorithm is to identify when a given solution achieves the desired sub-optimality with the required confidence, allowing the search to halt and return the incumbent solution. In this paper we propose two novel methods for identifying when a PAC search can halt. Unlike previous work, the new methods provided in this paper become more knowledgeable as the search progresses. This can speedup the search, since the search can halt earlier with the proposed methods and still keeping the desired PAC solution quality guarantees.Experimental results indeed show a substantial speedup of the search in comparison to the previous approach for PAC search.

show abstract

“…Our approach uses supervised learning of pattern database heuristics (Culberson and Schaeffer 1998), which are a common memory-based heuristic,and then use an ANN as a means of compression to reduce the storage required. Compression has been widely studied for memory-based heuristics (Felner et al 2007;Ball and Holte 2008;Breyer and Korf 2010;Goldenberg et al 2011), and is particularly important when the heuristics require memory that is larger than RAM (Döbbelin, Schütt, and Reinefeld 2013;Hu and Sturtevant 2019).…”

Section: Introductionmentioning

confidence: 99%

Optimal Search with Neural Networks: Challenges and Approaches

Chen

Mavrin

et al. 2022

SOCS

View full text Add to dashboard Cite

Work in machine learning has grown tremendously in the past years, but has had little to no impact on optimal search approaches. This paper looks at challenges in using deep learning as a part of optimal search, including what is feasible using current public frameworks, and what barriers exist for further adoption. The primary contribution of the paper is to show how to learn admissible heuristics through supervised learning from an existing heuristic. Several approaches are described, with the most successful approach being based on learning a heuristic as a classifier and then adjusting the quantile used with the classifier to ensure heuristic admissibility, which is required for optimal solutions. A secondary contribution is a description of the Batch A* algorithm, which can batch evaluations for more efficient use by the GPU. While ANNs can effectively learn heuristics that produce smaller search trees than alternate compression approaches, there still exists a time overhead when compared to efficient C++ implementations. This point of evaluation points out a challenge for future work.

show abstract

The Compressed Differential Heuristic

Cited by 5 publications

References 4 publications

Hierarchical Graph Traversal for Aggregate k Nearest Neighbors Search in Road Networks

Hierarchical Graph Traversal for Aggregate k Nearest Neighbors Search in Road Networks

Search-Aware Conditions for Probably Approximately Correct Heuristic Search

Optimal Search with Neural Networks: Challenges and Approaches

Contact Info

Product

Resources

About