Pith. sign in

REVIEW 4 major objections 5 minor 86 references

Graph Learning for Planning: The Story Thus Far and Open Challenges

T0 review · 4 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The paper argues that for planning, the best graph-learning recipe is a linear model on Weisfeiler-Lehman features trained to rank states instead of estimating costs to go.

desk verdict A useful, honest synthesis of the authors' own recent results, but the headline scalability claim rests on aggregate coverage rather than size-stratified evidence. read the letter →

arxiv 2412.02136 v1 pith:6I4L4ONZ submitted 2024-12-03 cs.AI

Dillon Z. Chen , Mingyu Hao , Sylvie Thiébaux , Felipe Trevizan This is my paper · ORCID
classification cs.AI
keywords graphlearningautomatedplanningWeisfeiler-LehmankernellearnedheuristicsrankingobjectivegeneralisationacrossobjectcountsGOOSEIPCtrack
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper sets out a position on graph learning for planning: after surveying graph representations, architectures, and training objectives, it argues the winning recipe is cheap, classical machine learning on Weisfeiler-Lehman graph features, trained with a ranking objective, rather than deep graph neural networks trained to regress optimal costs. It grounds this in a theory hierarchy showing which planning-task graph representations are more expressive under message-passing, and in experiments on the 2023 planning-competition learning track where the resulting heuristics guide greedy search to solve up to 619 of 900 problems, ahead of a strong baseline planner's 557. The practical stake is that learned planning heuristics can be built from small training tasks and scale to much larger test tasks, at a fraction of the training cost of deep models.

What carries the argument

The load-bearing mechanism is the Weisfeiler-Lehman (WL) colour-refinement algorithm, used as a graph kernel to turn a planning task's relational structure into a feature vector for a linear model; the WL algorithm is also the known expressivity ceiling for message-passing GNNs, which is why the paper views it as a sufficient feature extractor. The second mechanism is ranking as a constrained optimisation problem: rather than fitting costs to go, the learner fits a function that respects per-state inequalities derived from optimal plan traces, comparing a state against its parent and siblings. These two mechanisms combine in GOOSE, which learns domain knowledge from small training instances and applies it to test instances with many more objects.

What would settle it

Run the GOOSE pipeline on a domain where the skill needed to solve a task changes with object count—for example, a domain where small tasks are solved by a local rule but large tasks require global connectivity or counting. If the WL linear ranking heuristic's coverage falls below the GNN or cost-to-go baseline on such a domain under the same training-testing protocol, the claim that this recipe is consistently best is falsified.

Watch

Extended reading notes

Core claim

The paper's central claim is that the choice of training objective matters more than model size for learned planning heuristics: formulating heuristic learning as ranking states, with a constrained-optimisation objective that only preserves the inequalities needed by greedy best-first search, yields better guidance than regressing h* values. On the model side, the paper claims that linear models over Weisfeiler-Lehman colour-refinement features match or beat message-passing neural networks across classical and numeric planning domains while training orders of magnitude faster, and that grounded graph representations are strictly more expressive than lifted ones in the relevant sense. Assembled into the GOOSE framework—learn on small tasks, plan on large tasks—these choices make learned heuristics competitive with strong hand-built planners on the 2023 IPC learning track.

Load-bearing premise

The whole scaling story rests on the assumption that a domain's structure relevant to planning is stable as task size grows, so features learned from small problems remain the right features on much larger problems.

Editorial extensions

If this is right

  • New learned-heuristic systems for classical and numeric planning should be compared against WL-kernel linear models trained with ranking, not only against GNN baselines, because the paper shows the cheap model wins on coverage.
  • Since WL features with a linear model train orders of magnitude faster, the bottleneck in learning for planning shifts to label generation and data collection rather than model fitting.
  • Because grounded graph representations are more expressive than lifted ones, representation choice should be a first-class design decision in learning-for-planning systems, not an implementation detail.
  • Ranking objectives that preserve only comparison inequalities are a drop-in replacement for mean-squared-error cost-to-go in greedy best-first search; adopting them should improve coverage without changing the search algorithm.
  • On the IPC23LT classical benchmarks, a parallel portfolio of WL heuristics (619) exceeds LAMA (557), so learned heuristics with simple search are now a serious alternative to engineered satisficing planners.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The same WL-plus-ranking recipe may transfer to other search problems where cost-to-go labels are expensive but pairwise ordering information is cheap, such as theorem proving, program synthesis, or puzzle solving; the paper does not test this.
  • If grounded representations' extra expressivity is the reason for their advantage, then architectures that approximate higher-order WL refinement could close the remaining gap to strong planners without moving to full deep learning; this is an extrapolation from the paper's expressivity hierarchy.
  • The paper's setup assumes task-size scaling within a fixed domain; a natural challenge it does not address is cross-domain transfer, where a model is trained in one domain and evaluated in another, or where object-count scaling is not the only distribution shift.
  • Because GOOSE's training is so cheap, one could in principle use active learning or iterative re-training during search, updating the ranking function on the fly; the paper's data-collection discussion implies this but does not implement it.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 5 minor

Summary. This paper surveys graph learning for planning and argues for three main claims: (1) a taxonomy and expressive-power hierarchy of graph representations of planning tasks, (2) classical machine learning with Weisfeiler-Lehman graph features substantially outperforms GNNs for planning heuristics, and (3) formulating heuristic learning as a ranking problem outperforms cost-to-go regression. These claims are supported by experiments on the IPC23LT benchmarks, summarized in Figure 8, where a WL-based heuristic with parallel portfolio reaches coverage 619 versus LAMA's 557 in classical planning. The paper also introduces the GOOSE framework, reports training-time and parameter-count advantages of WL models over GNNs, and ends with five open challenges for the Learning for Planning field.

Significance. If the central claims hold, the paper makes a useful and somewhat contrarian contribution: cheap linear models over WL kernels can be more data-efficient, faster, and better than GNNs for planning heuristics, and the ranking objective is better aligned with GBFS than cost-to-go regression. The paper credits the ranking idea to prior independent work [GKL16, CEKP23] and evaluates on an external benchmark (IPC23LT), which are strengths. The open-challenges discussion, especially on expressivity and out-of-distribution generalization, is thoughtful. However, much of the material summarizes the authors' own prior papers, and the new synthesis does not yet provide the derivations, experimental details, or stratified analysis needed to verify the strongest comparative and transfer claims.

major comments (4)
  1. [§7.1, §7.3, Fig. 8] The load-bearing claim that the WL-based recipe transfers from small training tasks to the much larger test tasks of IPC23LT is not established by the aggregate coverage numbers in Fig. 8. Figure 7 shows that test tasks are often up to an order of magnitude larger than training tasks in object count (e.g., Blocksworld 488 vs 21, Floortile 1461 vs 35), and §8.II itself states that L4P is inherently out-of-distribution. Aggregate coverage over 900 problems could be dominated by near-training-size tasks while failing on the large tasks that motivate the framework. The authors should report coverage broken down by task-size buckets (per domain or globally) for hWL_rank, hWL_ptfl, and the GNN baselines, or otherwise show that WL features remain informative on the largest tasks.
  2. [§4.2, Fig. 4] The expressive-power hierarchy in Fig. 4 is presented as a set of results, but the paper gives no formal definitions, theorem statements, or proofs, nor does it point to specific theorems in the cited papers. In particular, 'more expressive', 'incomparable', and 'weaker notion of expressivity' are never defined in the text, and the distinction between incomparability with and without the h* difference requirement is unexplained. Since this hierarchy is one of the three contributions announced in the introduction, the authors should state the exact notion of distinguishability (e.g., which MPNN variant, which readout, which input features) and either provide the derivations or cite precise theorem numbers in [CTT24a, CTT24b].
  3. [§5, §7.1, §7.2, Fig. 5, Fig. 8] The comparison between WL models and GNNs is missing the experimental configuration needed to assess fairness and reproducibility. The paper reports no GNN architecture details (number of layers, hidden dimensions, normalization, pooling), no training hyperparameters (learning rate, optimizer, number of epochs), no number of seeds, and no variance or confidence intervals. The claim 'Classical ML consistently outperform deep learning' is a central takeaway (§7.3), and it depends on the GNN baselines being reasonably configured. The authors should provide these details and, if possible, report the distribution of coverage across seeds.
  4. [§6.3, §7.3, Fig. 8] The advantage of ranking over cost-to-go is motivated by examples (Fig. 6) and credited to [GKL16], but the paper does not provide a formal statement of when ranking enlarges the hypothesis space or why zero cost-to-go loss is insufficient. Moreover, the empirical evidence in classical planning is a difference of 18 problems (hWL_rank 520 vs hWL_cost 502), which may be within noise; no confidence intervals or per-domain breakdowns are given. The authors should either provide a formal argument for the ranking advantage or present per-domain results and significance measures.
minor comments (5)
  1. [§3.1] The notation [ [n] ] appears with a formatting artifact; if it denotes {1, ..., n}, use [n] or define it cleanly.
  2. [§4.2, Fig. 4] The caption mentions 'incomparable with no h* difference requirement', but this notion is never defined in the text; please define it.
  3. [§7.1] The abbreviation hWL_grid is listed but not explained; add a sentence describing the grid-search configuration and the range of hyperparameters searched.
  4. [§4.1] The sentence 'to the use of various different graphs and graph algorithms' contains a repeated 'to the use' and should be edited for readability.
  5. [Fig. 7] For several domains the difference between training and test sizes is hard to read on the log scale; consider adding numerical annotations or a table.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: central empirical claims are benchmark-based and the ranking idea is credited to independent prior work.

full rationale

This paper is a summary of the authors' prior work rather than a derivation in which predictions are obtained from fitted inputs. The central empirical claims in Sections 5 and 7.3 are comparisons of learned heuristics on the external IPC23LT benchmark, with coverage figures in Fig. 8; these results are falsifiable and are not entailed by the modeling definitions. The ranking formulation is explicitly credited to independent prior work (Garrett et al. [GKL16] and Chrestien et al. [CEKP23]), and the WL-feature motivation relies on established external results ([MRF+19, XHLJ19, SSVL+11]). The paper's self-citations, e.g., [CTT24b] for 'classical ML consistently outperform deep learning for symbolic planning', are to published, externally evaluated work rather than to an unverified premise or a forbidden uniqueness theorem. Section 8.II explicitly acknowledges that L4P is out-of-distribution and that standard generalization bounds do not straightforwardly apply; this is an honest open-challenge statement, not a hidden circular assumption. The GOOSE framework is an empirical pipeline, not a claim whose conclusion equals its definition. The strongest coverage numbers use per-domain configuration (580) or parallel portfolios (619), which are configurational upper bounds rather than single-model predictions; this is an evaluation-fairness concern, not circularity. The absence of size-stratified coverage is an evidential gap regarding scalability, not a step that reduces to its own inputs. Therefore, no circular step is present and the score is 0.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

The paper's central claims are empirical and comparative, not derived from a small set of first principles. The main implicit premises are transferability of learned features across object counts, adequacy of graph encodings, validity of WL-based features, and the benchmark metric. No new theoretical entities are introduced; GOOSE is a software framework with public code.

free parameters (4)
  • WL kernel iteration depth = not reported; grid-searched per domain for hWL_grid
    Controls colour-refinement granularity. The paper reports hWL_grid and hWL_ptfl configurations but does not give the selected values.
  • GNN architecture hyperparameters (layers, hidden dimensions) = not reported in this paper
    The hGNN baselines in Figure 8 depend on these choices; without sweeps or seeds, the comparison may not be representative.
  • Ranking loss margin or LP regularization = not reported in this paper
    The constrained-optimisation ranking formulation in Section 6 and [CT24a] requires margin or regularisation parameters, not stated in this summary.
  • Training label budget and training task sizes = not reported in this paper
    Learned coverage depends on how many h* labels and which small tasks were used. The paper refers to prior work but does not restate the budgets.
assumptions (4)
  • standard math MPNN expressivity is upper bounded by Weisfeiler-Lehman color refinement.
    Invoked in Sections 4 and 5 to justify comparing representations and to motivate WL kernels; based on [MRF+19, XHLJ19].
  • domain assumption The chosen graph encodings of planning tasks preserve enough information to distinguish tasks and guide search.
    Section 4 assumes planning tasks can be usefully encoded with object, predicate, and goal nodes or edges. Lossy encodings would invalidate the expressivity hierarchy and learned heuristics.
  • domain assumption Learning from small training tasks transfers to larger tasks in the same domain.
    Defines the L4P setup in Section 2 and underpins the IPC23LT experiments in Section 7, where test tasks have up to an order of magnitude more objects.
  • domain assumption Coverage on IPC23LT within a 30 minute, 8GB limit is a valid measure of planning performance.
    Section 7 uses coverage as the primary metric; all comparisons and takeaways are relative to this operationalization.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Graph Learning for Planning: The Story Thus Far and Open Challenges." pith.science (2026). https://pith.science/paper/6I4L4ONZ

@misc{pith2026241202136,
  author       = {Pith},
  title        = {Pith review of: Graph Learning for Planning: The Story Thus Far and Open Challenges},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/6I4L4ONZ}},
  note         = {Machine review of arXiv:2412.02136}
}
read the original abstract

Graph learning is naturally well suited for use in planning due to its ability to exploit relational structures exhibited in planning domains and to take as input planning instances with arbitrary number of objects. In this paper, we study the usage of graph learning for planning thus far by studying the theoretical and empirical effects on learning and planning performance of (1) graph representations of planning tasks, (2) graph learning architectures, and (3) optimisation formulations for learning. Our studies accumulate in the GOOSE framework which learns domain knowledge from small planning tasks in order to scale up to much larger planning tasks. In this paper, we also highlight and propose the 5 open challenges in the general Learning for Planning field that we believe need to be addressed for advancing the state-of-the-art.

Figures

Figures reproduced from arXiv: 2412.02136 by the authors.

Figure 1
Figure 1. Typical graph learning for planning pipeline. (1) [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Visualisations of generalisation setups for planning [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 5
Figure 5. Time to train after data preprocessing in log scale [PITH_FULL_IMAGE:figures/full_fig_p004_5.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Expressivity hierarchy of graph representations of [PITH_FULL_IMAGE:figures/full_fig_p004_4.png]
Figure 6
Figure 6. Figure 6: Visualisations of heuristics that achieve zero [PITH_FULL_IMAGE:figures/full_fig_p005_6.png]
Figure 7
Figure 7. Figure 7: Sizes of training and testing tasks in the IPC23LT. [PITH_FULL_IMAGE:figures/full_fig_p006_7.png]
Figure 8
Figure 8. Figure 8: Coverage of classic/numeric (purple) and learning [PITH_FULL_IMAGE:figures/full_fig_p006_8.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

86 extracted references · 76 canonical work pages

  1. [1]

    , " * write output.state after.block = add.period write newline

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint howpublished institution isbn journal key month note number organization pages publisher school series title type volume year label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence after.sentence after.block FUNCTION init.state.consts #0 'before.a...

  2. [2]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in capitalize " " * FUNCT...

  3. [3]

    The surprising power of graph neural networks with random node initialization

    Ralph Abboud, \.I smail \.I lkan Ceylan, Martin Grohe, and Thomas Lukasiewicz. The surprising power of graph neural networks with random node initialization. In IJCAI , 2021

    2021

  4. [4]

    Improving subgraph-gnns via edge-level ego-network encodings

    Nurudin Alvarez - Gonzalez, Andreas Kaltenbrunner, and Vicen c G \' o mez. Improving subgraph-gnns via edge-level ego-network encodings. Trans. Mach. Learn. Res. , 2024, 2024

    2024

  5. [5]

    Avrim Blum and Merrick L. Furst. Fast planning through planning graph analysis. Artif. Intell. , 90(1-2):281--300, 1997

    1997

  6. [6]

    Bronstein

    Giorgos Bouritsas, Fabrizio Frasca, Stefanos Zafeiriou, and Michael M. Bronstein. Improving graph neural network expressivity via subgraph isomorphism counting. IEEE Trans. Pattern Anal. Mach. Intell. , 45(1), 2023

    2023

  7. [7]

    General policies, subgoal structure, and planning width

    Blai Bonet and Hector Geffner. General policies, subgoal structure, and planning width. J. Artif. Intell. Res. , 80:475--516, 2024

    work page 2024

  8. [8]

    Breaking the limits of message passing graph neural networks

    Muhammet Balcilar, Pierre H \' e roux, Benoit Ga \" u z \` e re, Pascal Vasseur, S \' e bastien Adam, and Paul Honeine. Breaking the limits of message passing graph neural networks. In ICML , 2021

    work page 2021

Show all 86 references
  1. [9]

    Kostylev, Mika \" e l Monet, Jorge P \' e rez, Juan L

    Pablo Barcel \' o , Egor V. Kostylev, Mika \" e l Monet, Jorge P \' e rez, Juan L. Reutter, and Juan Pablo Silva. The logical expressiveness of graph neural networks. In ICLR , 2020

    2020

  2. [10]

    Bartlett and Shahar Mendelson

    Peter L. Bartlett and Shahar Mendelson. Rademacher and gaussian complexities: Risk bounds and structural results. In COLT/EuroCOLT , 2001

    2001

  3. [11]

    The computational complexity of propositional STRIPS planning

    Tom Bylander. The computational complexity of propositional STRIPS planning. Artif. Intell. , 69(1-2):165--204, 1994

    1994

  4. [12]

    A review of generalized planning

    Sergio Jim \' e nez Celorrio, Javier Segovia Aguas, and Anders Jonsson. A review of generalized planning. Knowl. Eng. Rev. , 34:e5, 2019

    2019

  5. [13]

    Muggleton

    Andrew Cropper, Sebastijan Dumancic, Richard Evans, and Stephen H. Muggleton. Inductive logic programming at 30. Mach. Learn. , 111(1):147--172, 2022

    2022

  6. [14]

    A review of machine learning for automated planning

    Sergio Jim \' e nez Celorrio, Tom \' a s de la Rosa, Susana Fern \' a ndez, Fernando Fern \' a ndez, and Daniel Borrajo. A review of machine learning for automated planning. Knowl. Eng. Rev. , 27(4):433--467, 2012

    2012

  7. [15]

    Optimize planning heuristics to rank, not to estimate cost-to-goal

    Leah Chrestien, Stefan Edelkamp, Anton \' n Komenda, and Tom \' a s Pevn \' y . Optimize planning heuristics to rank, not to estimate cost-to-goal. In NeurIPS , 2023

    2023

  8. [16]

    Deep learning for generalised planning with background knowledge

    Dillon Ze Chen, Rostislav Hor c \' k, and Gustav S \' r. Deep learning for generalised planning with background knowledge. CoRR , abs/2410.07923, 2024

    2024 arXiv

  9. [17]

    Chen and Sylvie Thi \' e baux

    Dillon Z. Chen and Sylvie Thi \' e baux. Graph learning for numeric planning. In NeurIPS , 2024

    2024

  10. [18]

    Chen and Sylvie Thi \' e baux

    Dillon Z. Chen and Sylvie Thi \' e baux. Novelty heuristics, multi-queue search, and portfolios for numeric planning. In SOCS , 2024

    2024

  11. [19]

    Chen, Sylvie Thi \' e baux, and Felipe Trevizan

    Dillon Z. Chen, Sylvie Thi \' e baux, and Felipe Trevizan. Learning domain-independent heuristics for grounded and lifted planning. In AAAI , 2024

    2024

  12. [20]

    Chen, Felipe Trevizan, and Sylvie Thi \' e baux

    Dillon Z. Chen, Felipe Trevizan, and Sylvie Thi \' e baux. Return to tradition: Learning reliable heuristics with classical machine learning. In ICAPS , 2024

    2024

  13. [21]

    Higher-dimensional potential heuristics: Lower bound criterion and connection to correlation complexity

    Simon Dold and Malte Helmert. Higher-dimensional potential heuristics: Lower bound criterion and connection to correlation complexity. In ICAPS , 2024

    2024

  14. [22]

    Novelty vs

    Simon Dold and Malte Helmert. Novelty vs. potential heuristics: A comparison of hardness measures for satisficing planning. In AAAI , 2024

    2024

  15. [23]

    Neural logic machines

    Honghua Dong, Jiayuan Mao, Tian Lin, Chong Wang, Lihong Li, and Denny Zhou. Neural logic machines. In ICLR , 2019

    2019

  16. [24]

    Equivalence-based abstractions for learning general policies

    Dominik Drexler, Simon St hlberg, Blai Bonet, and Hector Geffner. Equivalence-based abstractions for learning general policies. In KR , 2024

    2024

  17. [25]

    Expressing and exploiting subgoal structure in classical planning using sketches

    Dominik Drexler, Jendrik Seipp, and Hector Geffner. Expressing and exploiting subgoal structure in classical planning using sketches. J. Artif. Intell. Res. , 80, 2024

    2024

  18. [26]

    How powerful are k-hop message passing graph neural networks

    Jiarui Feng, Yixin Chen, Fuhai Li, Anindya Sarkar, and Muhan Zhang. How powerful are k-hop message passing graph neural networks. In NeurIPS , 2022

    2022

  19. [27]

    Neural network heuristic functions for classical planning: Bootstrapping and comparison to other methods

    Patrick Ferber, Florian Gei er, Felipe Trevizan, Malte Helmert, and J \" o rg Hoffmann. Neural network heuristic functions for classical planning: Bootstrapping and comparison to other methods. In ICAPS , 2022

    2022

  20. [28]

    PDDL2.1: an extension to PDDL for expressing temporal planning domains

    Maria Fox and Derek Long. PDDL2.1: an extension to PDDL for expressing temporal planning domains. J. Artif. Intell. Res. , 20:61--124, 2003

    2003

  21. [29]

    A Concise Introduction to Models and Methods for Automated Planning

    Hector Geffner and Blai Bonet. A Concise Introduction to Models and Methods for Automated Planning . Synthesis Lectures on Artificial Intelligence and Machine Learning. Morgan & Claypool Publishers, 2013

    2013

  22. [30]

    Learning to rank for synthesizing planning heuristics

    Caelan Reed Garrett, Leslie Pack Kaelbling, and Tom \' a s Lozano - P \' e rez. Learning to rank for synthesizing planning heuristics. In IJCAI , 2016

    2016

  23. [31]

    Formal representations of classical planning domains

    Claudia Grundke, Gabriele R \" o ger, and Malte Helmert. Formal representations of classical planning domains. In ICAPS , 2024

    2024

  24. [32]

    The logic of graph neural networks

    Martin Grohe. The logic of graph neural networks. In LICS , 2021

    2021

  25. [33]

    LPG: A planner based on local search for planning graphs with action costs

    Alfonso Gerevini and Ivan Serina. LPG: A planner based on local search for planning graphs with action costs. In AAAI , 2002

    2002

  26. [34]

    Schoenholz, Patrick F

    Justin Gilmer, Samuel S. Schoenholz, Patrick F. Riley, Oriol Vinyals, and George E. Dahl. Neural message passing for quantum chemistry. In ICML , 2017

    2017

  27. [35]

    Exploiting first-order regression in inductive policy selection

    Charles Gretton and Sylvie Thi \' e baux. Exploiting first-order regression in inductive policy selection. In UAI , 2004

    2004

  28. [36]

    Landmarks, critical paths and abstractions: What's the difference anyway? In ICAPS , 2009

    Malte Helmert and Carmel Domshlak. Landmarks, critical paths and abstractions: What's the difference anyway? In ICAPS , 2009

    2009

  29. [37]

    A planning heuristic based on causal graph analysis

    Malte Helmert. A planning heuristic based on causal graph analysis. In ICAPS , 2004

    2004

  30. [38]

    The FF planning system: Fast plan generation through heuristic search

    J \" o rg Hoffmann and Bernhard Nebel. The FF planning system: Fast plan generation through heuristic search. J. Artif. Intell. Res. , 14:253--302, 2001

    2001

  31. [39]

    Expressiveness of graph neural networks in planning domains

    Rostislav Hor c \' k and Gustav S \' r. Expressiveness of graph neural networks in planning domains. In ICAPS , 2024

    2024

  32. [40]

    Guiding GBFS through learned pairwise rankings

    Mingyu Hao, Felipe Trevizan, Sylvie Thiébaux, Patrick Ferber, and Jörg Hoffmann. Guiding GBFS through learned pairwise rankings. In IJCAI , 2024

    2024

  33. [41]

    McIlraith

    Le \' o n Illanes and Sheila A. McIlraith. Generalized planning via abstraction: Arbitrary numbers of objects. In AAAI , 2019

    2019

  34. [42]

    Relational queries computable in polynomial time (extended abstract)

    Neil Immerman. Relational queries computable in polynomial time (extended abstract). In STOC , 1982

    1982

  35. [43]

    Learning action strategies for planning domains

    Roni Khardon. Learning action strategies for planning domains. Artif. Intell. , 113(1-2):125--148, 1999

    1999

  36. [44]

    Kriege, Fredrik D

    Nils M. Kriege, Fredrik D. Johansson, and Christopher Morris. A survey on graph kernels. Appl. Netw. Sci. , 5(1):6, 2020

    2020

  37. [45]

    Learning generalized relational heuristic networks for model-agnostic planning

    Rushang Karia and Siddharth Srivastava. Learning generalized relational heuristic networks for model-agnostic planning. In AAAI , 2021

    2021

  38. [46]

    Yann LeCun, Yoshua Bengio, and Geoffrey E. Hinton. Deep learning. Nat. , 521(7553):436--444, 2015

    2015

  39. [47]

    Width and serialization of classical planning problems

    Nir Lipovetzky and Hector Geffner. Width and serialization of classical planning problems. In ECAI , 2012

    2012

  40. [48]

    Bronstein, Martin Grohe, and Stefanie Jegelka

    Christopher Morris, Fabrizio Frasca, Nadav Dym, Haggai Maron, \.I smail \.I lkan Ceylan, Ron Levie, Derek Lim, Michael M. Bronstein, Martin Grohe, and Stefanie Jegelka. Position: Future directions in the theory of graph machine learning. In ICML , 2024

    2024

  41. [49]

    Online planner selection with graph neural networks and adaptive scheduling

    Tengfei Ma, Patrick Ferber, Siyu Huo, Jie Chen, and Michael Katz. Online planner selection with graph neural networks and adaptive scheduling. In AAAI , 2020

    2020

  42. [50]

    Tenenbaum, and Leslie Pack Kaelbling

    Jiayuan Mao, Tom \' a s Lozano - P \' e rez, Joshua B. Tenenbaum, and Leslie Pack Kaelbling. What planning problems can A relational neural network solve? In NeurIPS , 2023

    2023

  43. [51]

    Hamilton, Jan Eric Lenssen, Gaurav Rattan, and Martin Grohe

    Christopher Morris, Martin Ritzert, Matthias Fey, William L. Hamilton, Jan Eric Lenssen, Gaurav Rattan, and Martin Grohe. Weisfeiler and leman go neural: Higher-order graph neural networks. In AAAI , 2019

    2019

  44. [52]

    Speqnets: Sparsity-aware permutation-equivariant graph networks

    Christopher Morris, Gaurav Rattan, Sandra Kiefer, and Siamak Ravanbakhsh. Speqnets: Sparsity-aware permutation-equivariant graph networks. In ICML , 2022

    2022

  45. [53]

    Weisfeiler and leman go sparse: Towards scalable higher-order graph embeddings

    Christopher Morris, Gaurav Rattan, and Petra Mutzel. Weisfeiler and leman go sparse: Towards scalable higher-order graph embeddings. In NeurIPS , 2020

    2020

  46. [54]

    On using admissible bounds for learning forward search heuristics

    Carlos N \' u \ n ez - Molina, Masataro Asai, Pablo Mesejo, and Juan Fern \' a ndez - Olivares. On using admissible bounds for learning forward search heuristics. In IJCAI , 2024

    2024

  47. [55]

    Laurent Orseau and Levi H. S. Lelis. Policy-guided heuristic search with guarantees. In AAAI , 2021

    2021

  48. [56]

    Rosenschein

    Nir Pochter, Aviv Zohar, and Jeffrey S. Rosenschein. Exploiting problem symmetries in state-based planners. In AAAI , 2011

    2011

  49. [57]

    The LAMA planner: Guiding cost-based anytime planning with landmarks

    Silvia Richter and Matthias Westphal. The LAMA planner: Guiding cost-based anytime planning with landmarks. J. Artif. Intell. Res. , 39:127--177, 2010

    2010

  50. [58]

    Sutton and Andrew G

    Richard S. Sutton and Andrew G. Barto. Reinforcement learning - an introduction . Adaptive computation and machine learning. MIT Press, 1998

    1998

  51. [59]

    Learning general optimal policies with graph neural networks: Expressive power, transparency, and limits

    Simon St hlberg, Blai Bonet, and Hector Geffner. Learning general optimal policies with graph neural networks: Expressive power, transparency, and limits. In ICAPS , 2022

    2022

  52. [60]

    Learning general policies with policy gradient methods

    Simon St hlberg, Blai Bonet, and Hector Geffner. Learning general policies with policy gradient methods. In KR , 2023

    2023

  53. [61]

    Learning general policies for classical planning domains: Getting beyond c\( _ 2 \)

    Simon St hlberg, Blai Bonet, and Hector Geffner. Learning general policies for classical planning domains: Getting beyond c\( _ 2 \). CoRR , abs/2403.11734, 2024

    2024 arXiv

  54. [62]

    Tenenbaum, Tom \' a s Lozano - P \' e rez, and Leslie Pack Kaelbling

    Tom Silver, Rohan Chitnis, Aidan Curtis, Joshua B. Tenenbaum, Tom \' a s Lozano - P \' e rez, and Leslie Pack Kaelbling. Planning with learned object importance in large problem instances using graph neural networks. In AAAI , 2021

    2021

  55. [63]

    Heuristics and symmetries in classical planning

    Alexander Shleyfman, Michael Katz, Malte Helmert, Silvan Sievers, and Martin Wehrle. Heuristics and symmetries in classical planning. In AAAI , 2015

    2015

  56. [64]

    Correlation complexity of classical planning domains

    Jendrik Seipp, Florian Pommerening, Gabriele R \" o ger, and Malte Helmert. Correlation complexity of classical planning domains. In IJCAI , 2016

    2016

  57. [65]

    Foundations and Applications of Generalized Planning

    Siddharth Srivastava. Foundations and Applications of Generalized Planning . PhD thesis, University of Massachusetts Amherst, 2010

    2010

  58. [66]

    Theoretical foundations for structural symmetries of lifted PDDL tasks

    Silvan Sievers, Gabriele R \" o ger, Martin Wehrle, and Michael Katz. Theoretical foundations for structural symmetries of lifted PDDL tasks. In ICAPS , 2019

    2019

  59. [67]

    Search-guidance mechanisms for numeric planning through subgoaling relaxation

    Enrico Scala, Alessandro Saetti, Ivan Serina, and Alfonso Emilio Gerevini. Search-guidance mechanisms for numeric planning through subgoaling relaxation. In ICAPS , 2020

    2020

  60. [68]

    Weisfeiler-lehman graph kernels

    Nino Shervashidze, Pascal Schweitzer, Erik Jan Van Leeuwen, Kurt Mehlhorn, and Karsten M Borgwardt. Weisfeiler-lehman graph kernels. J. Mach. Learn. Res. , 12:2539--2561, 2011

    2011

  61. [69]

    Trevizan, Sam Toyer, Sylvie Thi \' e baux, and Lexing Xie

    William Shen, Felipe W. Trevizan, Sam Toyer, Sylvie Thi \' e baux, and Lexing Xie. Guiding search with generalized policies for probabilistic planning. In SOCS , 2019

    2019

  62. [70]

    L earning D omain- I ndependent P lanning H euristics with H ypergraph N etworks

    William Shen, Felipe Trevizan, and Sylvie Thi \'e baux. L earning D omain- I ndependent P lanning H euristics with H ypergraph N etworks. In ICAPS , 2020

    2020

  63. [71]

    The 2023 international planning competition

    Ayal Taitler, Ron Alford, Joan Espasa, Gregor Behnke, Daniel Fiser, Michael Gimelfarb, Florian Pommerening, Scott Sanner, Enrico Scala, Dominik Schreiber, Javier Segovia - Aguas, and Jendrik Seipp. The 2023 international planning competition. AI Mag. , 45(2):280--296, 2024

    2023

  64. [72]

    Trevizan, Sylvie Thi \' e baux, and Lexing Xie

    Sam Toyer, Felipe W. Trevizan, Sylvie Thi \' e baux, and Lexing Xie. Action schema networks: Generalised policies with deep learning. In AAAI , 2018

    2018

  65. [73]

    Asnets: Deep learning for generalised planning

    Sam Toyer, Sylvie Thi \' e baux, Felipe Trevizan, and Lexing Xie. Asnets: Deep learning for generalised planning. J. Artif. Intell. Res. , 68:1--68, 2020

    2020

  66. [74]

    Statistical learning theory

    Vladimir Vapnik. Statistical learning theory . Wiley, 1998

    1998

  67. [75]

    Moshe Y. Vardi. The complexity of relational query languages (extended abstract). In STOC , 1982

    1982

  68. [76]

    Planbench: An extensible benchmark for evaluating large language models on planning and reasoning about change

    Karthik Valmeekam, Matthew Marquez, Alberto Olmo Hernandez, Sarath Sreedharan, and Subbarao Kambhampati. Planbench: An extensible benchmark for evaluating large language models on planning and reasoning about change. In NeurIPS , 2023

    2023

  69. [77]

    On the planning abilities of large language models - A critical investigation

    Karthik Valmeekam, Matthew Marquez, Sarath Sreedharan, and Subbarao Kambhampati. On the planning abilities of large language models - A critical investigation. In NeurIPS , 2023

    2023

  70. [78]

    Llms still can't plan; can lrms? A preliminary evaluation of openai's o1 on planbench

    Karthik Valmeekam, Kaya Stechly, and Subbarao Kambhampati. Llms still can't plan; can lrms? A preliminary evaluation of openai's o1 on planbench. CoRR , abs/2409.13373, 2024

    2024 arXiv

  71. [79]

    N -wl: A new hierarchy of expressivity for graph neural networks

    Qing Wang, Dillon Ze Chen, Asiri Wijesinghe, Shouheng Li, and Muhammad Farhan. N -wl: A new hierarchy of expressivity for graph neural networks. In ICLR , 2023

    2023

  72. [80]

    Wang and Sylvie Thi \' e baux

    Ryan X. Wang and Sylvie Thi \' e baux. Learning generalised policies for numeric planning. In ICAPS , 2024

    2024

  73. [81]

    How powerful are graph neural networks? In ICLR , 2019

    Keyulu Xu, Weihua Hu, Jure Leskovec, and Stefanie Jegelka. How powerful are graph neural networks? In ICLR , 2019

    2019

  74. [82]

    From stars to subgraphs: Uplifting any GNN with local structure awareness

    Lingxiao Zhao, Wei Jin, Leman Akoglu, and Neil Shah. From stars to subgraphs: Uplifting any GNN with local structure awareness. In ICLR , 2022

    2022

  75. [83]

    A practical, progressively-expressive GNN

    Lingxiao Zhao, Neil Shah, and Leman Akoglu. A practical, progressively-expressive GNN . In NeurIPS , 2022

    2022

  76. [84]

    @esa ( ) , n @biblabelnum##1 ##1

    \@ifclassloaded aguplus natbib The aguplus class already includes natbib coding, so you should not add it explicitly Type <Return> for now, but then later remove the command natbib from the document \@ifclassloaded nlinproc natbib The nlinproc class already includes natbib cod...

  77. [85]

    @stdbsttrue NAT@ctr \@lbibitem[ NAT@ctr ] \@lbibitem[#1]#2 \@extra@b@citeb \@ifundefined br@#2\@extra@b@citeb \@namedef br@#2 \@nameuse br@#2\@extra@b@citeb \@ifundefined b@#2\@extra@b@citeb @num @parse #2 [ @natanchorstart #2\@extra@b@citeb \@biblabel @num @natanchorend] @ifc...

  78. [86]

    THr_v e 1G :۲Q;;Z ^׎ h&@i pWzf D p@cgYd >i 3 )b!VJV 6i@ '1`v0 :x 3 7_#:G['z ;/` Yf97' Pd VU`T r#. 2/BP ùĵC0ҼC b78 @SK

    @open @close @open @close and [1] URL: #1 \@ifundefined chapter * \@mkboth \@ifundefined NAT@sectionbib * \@mkboth * \@mkboth\@gobbletwo \@ifclassloaded amsart * \@ifclassloaded amsbook * \@ifundefined bib@heading @heading NAT@ctr thebibliography [1] @ \@biblabel NAT@ctr \@bib...

    1952

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.