Pith. sign in

REVIEW 3 major objections 4 minor 85 references

ASNets: Deep Learning for Generalised Planning

T0 review · 3 major / 4 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read By tying network weights to planning-domain schemas, the paper claims a policy learned on small problems transfers to much larger ones.

desk verdict A solid, transparent empirical paper on learning generalised planning policies; the headline Blocksworld result is real but rests on a tuning choice and an unproven transfer assumption. read the letter →

arxiv 1908.01362 v2 pith:XHGOEB2D submitted 2019-08-04 cs.AI

Sam Toyer , Felipe Trevizan , Sylvie Thiébaux , Lexing Xie This is my paper · ORCID
classification cs.AI
keywords generalisedplanningactionschemanetworksdeeplearningPPDDLimitationreactivepoliciesBlocksworldprobabilistic
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 claims that a single neural network with weights tied to the schemas of a planning domain can learn a generalised reactive policy by imitating a traditional planner on a few small problems, and that this policy then solves much larger instances from the same domain. The load-bearing evidence is an experiment in which training on 50 Blocksworld problems with 8 to 10 blocks yielded a policy that solved all 18,300 test instances with 18 to 50 blocks. Across seven probabilistic and deterministic domains, the network matches or beats heuristic-search planners in cumulative coverage once its two-hour training period is accounted for, solving several instances the planners could not solve within three hours. The paper also reports that sparsity regularisation can shrink a trained policy to a small set of human-readable equations, and it explicitly notes domains where this transfer recipe fails, including n-Puzzle and Sokoban.

What carries the argument

The central mechanism is the action-proposition relatedness graph: each action module receives, at each layer, the hidden representations of propositions appearing at unique positions in its action schema, while each proposition module pools over related actions grouped by schema and position. Weight tying across modules instantiated from the same action schema or predicate makes the parameter set $ heta$ independent of problem size, and first-layer inputs add proposition truth values, goal indicators, action applicability, LM-cut landmark indicators, and action counts. The output layer uses a masked softmax to produce a distribution over applicable actions. This construction is what lets a fixed set of weights act as a generalised policy over an entire lifted-SSP domain.

What would settle it

Train an ASNet on Blocksworld problems whose goal towers contain misplaced blocks only near the top, then test on an instance where the only misplaced block sits at the bottom of a tall tower; if the network's actions loop or fail there, the transfer assumption is violated. A more direct check is the paper's own receptive-field experiment: a fixed-depth ASNet should fail on a chain-length problem longer than its number of proposition layers.

Watch

Extended reading notes

Core claim

An ASNet is a neural network instantiated over the graph of ground actions and propositions of a (P)PDDL problem, with shared weights that depend only on action schemas and predicates, so the same parameter set applies to every instance of a domain. Trained by imitation learning against the value estimates of a heuristic-search teacher on small problems, the resulting reactive policy is claimed to transfer to much larger instances, with the strongest result being 18,300 of 18,300 Blocksworld test instances solved. The paper presents the architecture's position-sensitive pooling, skip connections, and heuristic input features as improvements that make such transfer possible, while acknowledging structural limits: no support for quantifiers or arbitrary goal formulae, a receptive-field limit on chains of reasoning, and failure to produce reliable policies for n-Puzzle and Sokoban.

Load-bearing premise

The small training problems are representative of the larger test problems, so the structure needed for a general policy is present in the training set; the paper offers strong empirical evidence in some domains but no formal transfer guarantee.

Editorial extensions

If this is right

  • A policy trained on small instances can solve problems that heuristic-search planners cannot solve within a three-hour budget, as demonstrated on Blocksworld, CosaNostra Pizza, Triangle Tireworld, and other domains.
  • The same architecture can handle both probabilistic and deterministic PPDDL domains because probabilistic semantics only enter through training data; the network itself reads the same relational structure.
  • Heuristic input features such as LM-cut landmarks and action-count history are essential in several domains, and removing them sharply reduces coverage.
  • The receptive-field limit implies that a fixed-depth network cannot generalise across domains where correct actions require reasoning over arbitrarily long relatedness chains, as shown by the paper's own chain-length experiments.
  • Sparsity regularisation can reduce a policy to very few nonzero parameters, producing a lifted policy that a human can read and check against the domain definition.

Reading between the lines

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

  • An implication the author leaves implicit is that the practical bottleneck for this approach is training-set design: to generalise, the small training problems must contain the rare structures (deep towers, dead-end configurations, long chains) that will appear in larger instances, and the paper's failures on n-Puzzle and Sokoban illustrate that this is not automatic.
  • A natural extension would be to combine ASNet policies with search at test time, using the learned policy to guide rather than replace a planner; the paper mentions this direction but does not evaluate it.
  • Because the sparse Triangle Tireworld policy reduces to eight nonzero weights, a further step would be to verify such policies formally against a specification of the domain, turning learned control knowledge into a checked artefact rather than an empirical claim.
  • The ASNet recipe of schema-derived graph structure plus shared weights and heuristic inputs could plausibly transfer to numeric and concurrent planning domains, though the paper leaves those extensions open.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 4 minor

Summary. The paper presents Action Schema Networks (ASNets), a graph-convolutional architecture for learning generalized policies in (P)PDDL planning domains. The network is structured by the relatedness of ground actions and propositions, with weight sharing across the same lifted action schemas and predicates, so that a fixed set of parameters can be applied to problems of arbitrary size. Policies are trained by imitation learning from a teacher planner on a handful of small instances, then executed greedily on much larger held-out instances. The paper extends the earlier ASNet architecture with position-aware pooling and skip connections, adds LM-cut landmark and action-count inputs to mitigate the receptive-field limitation, and reports experiments on seven probabilistic and deterministic domains, an extended Blocksworld evaluation of 18,300 instances, ablations, and an interpretability study using L1-regularized sparse policies.

Significance. If the results hold, this is an important contribution to generalized planning and to structured deep learning for planning. The architecture is a clean and general way to obtain size-invariant neural policies with strong inductive biases, the comparison against strong non-learning baselines is extensive and fairly presented, and the authors honestly report domains where the approach fails (n-Puzzle, Sokoban, Parking, Thoughtful). The paper ships code, provides detailed cost/coverage tables, and demonstrates a genuinely interpretable policy for Triangle Tireworld. The main limitations are the absence of a formal generalization guarantee and the reliance on hand-engineered heuristic input features, both of which the paper acknowledges; the empirical strength of the flagship 18,300-instance result is the key remaining question.

major comments (3)
  1. [Section 5.5] The headline claim that the trained policy solves all 18,300 test instances is reported after modifying the hyperparameters and training setup from Section 5.1 (20 instead of 16 channels, 50 instead of 25 training problems, six instead of two hours, dropout 0.3 instead of 0.1). No seed variance or separate validation set is reported, and the text does not describe a model-selection protocol that would rule out using the extended test set to choose these settings. As a result, the 100% coverage figure is best interpreted as an optimistic point estimate rather than a reliable generalization rate. Please report coverage over multiple training seeds and clarify the selection protocol, or soften the claim accordingly.
  2. [Section 3.5] The receptive-field limitation means the architecture alone cannot solve large instances in most domains; the LM-cut landmark features and action-count features are essential, as shown by the 'no h.' rows in Table 1 and the receptive-field experiments in Appendix C. This does not invalidate the method, but it does mean the claimed generalization is a property of the ASNet-plus-heuristic-features system rather than of the learned weights alone. The abstract and conclusion should qualify the claim, for example by stating that the policy is learned on top of a domain-independent landmark heuristic.
  3. [Section 5.1.1] The hyperparameters were selected by automated tuning and manual interpolation to maximize coverage on the benchmark problems, and the same benchmark suite is later used to evaluate the final system. This creates a selection bias that is not quantified. Please report a sensitivity analysis over the flagged hyperparameters (dropout, L2 coefficient, learning rate, hidden dimension) or a nested evaluation with a separate validation set, so that the comparison against untuned off-the-shelf baselines is not systematically disadvantaged.
minor comments (4)
  1. [Figure 4 caption] The caption says 'four drive actions related to at(shakey, kitchen)', but the text in Section 3.3 uses at(shakey, hall) for this example; the caption should be corrected to match the text.
  2. [Section 6.2] The sentence 'the ELU is simply the identify function on [0,∞)' contains a typo: 'identify' should be 'identity'.
  3. [Table 2] The column layout of Table 2 is confusing; please add explicit labels for the instance groups (e.g., '18 blocks, constrained towers', '25 blocks, random towers') so that the row structure matches the 18,300 total.
  4. [Section 4.1] The phrase 'containing structural elements representative of those found in larger problems' is vague; a brief concrete characterization (e.g., range of tower counts, goal configurations) would help readers understand the transfer assumption.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the paper is an empirical imitation-learning study evaluated on held-out instances, with no prediction that reduces to a fitted parameter or self-citation chain.

full rationale

The paper's central claim is that ASNets can learn a generalised policy by imitating a teacher planner on small instances and then solve larger held-out instances. This is an empirical claim, not a derivation: the network weights are learned from teacher-labelled states on training problems, and performance is measured on separate test problems from the same domains (Sections 4.1 and 5). There is no step in which a fitted parameter is renamed as a prediction. The architecture is described in full (Section 3), and the extensions over the prior ASNets paper (Toyer et al. 2018) are presented with their own ablations and analyses, so the self-citation is background rather than load-bearing. The heuristic inputs derived from LM-cut landmarks are not circular because the paper explicitly shows that baseline planners using LM-cut alone do not achieve ASNet coverage (Section 5.3.3, 'Are LM-cut heuristic values sufficient to trivially solve our test tasks?'). The extended Blocksworld experiment is a strong empirical result, and the paper transparently reports that the hyperparameters were modified for that experiment; this is a robustness concern about tuning and validation, not a circular reduction of the claim to its inputs. The paper also openly reports domains where the method fails (n-Puzzle, Sokoban, Parking, Thoughtful in Section 5.3.3), which is consistent with an honest empirical evaluation rather than a circular argument. Overall, no circular step can be exhibited from the text, so the appropriate score is 0.

Assumptions & free parameters 7 free parameters · 5 assumptions · 0 invented entities

The central claims are empirical, so the ledger mainly lists the hyperparameters that were tuned on the benchmark problems and the domain assumptions about transfer and teacher quality. No new physical entities are introduced.

free parameters (7)
  • hidden dimension dh = 16 (20 for extended Blocksworld)
    The width of the action and proposition modules; chosen via automated tuning and manual interpolation to maximize coverage on benchmark problems (Section 5.1.1).
  • number of proposition layers L = 2 (ablated 1 and 3)
    Depth of the network; selected based on tuning; affects receptive field and expressiveness.
  • dropout probability = 0.1 (0.3 for extended Blocksworld)
    Regularization; tuned to trade off overfitting and coverage.
  • L2 regularization coefficient = 2e-4
    Weight penalty in the training objective; tuned.
  • learning rate = 1e-3 (decayed for sparse policies)
    Adam optimizer learning rate; tuned.
  • L1 regularization coefficient (sparse policies) = 1e-2 (CosaNostra 0.02)
    Sparsity penalty used to obtain interpretable policies in Section 6; chosen by experiment.
  • training time cap = 2 hours (6 hours extended, 8 hours sparse)
    Computational budget; affects final weights and coverage.
assumptions (5)
  • domain assumption The relatedness graph derived from PPDDL action schemas captures the structural information needed for a generalised policy.
    Section 3.1 defines relatedness; the architecture and weight sharing assume this graph is sufficient to express and transfer policies across problems.
  • domain assumption The teacher planner produces Q-values that identify optimal or near-optimal actions in the training states.
    Section 4.1 labels actions by comparing Q-values; if the teacher is suboptimal or not converged, the policy inherits the bias, as the paper itself notes.
  • domain assumption The training problem distribution is representative of the test distribution.
    Section 4.1 and 5.2 assume small training problems contain the structural motifs needed for large instances; no formal guarantee is given.
  • standard math The ELU activation function is strictly increasing.
    Used in Appendix D to justify comparisons of activations in the sparse CosaNostra policy verification.
  • ad hoc to paper LM-cut landmarks and action counts provide enough information to overcome the receptive field limitation.
    Section 3.5 introduces these heuristic input features; their sufficiency is established only empirically through ablations.

how reviews work

0 comments
Cite this review

Pith. "Pith review of ASNets: Deep Learning for Generalised Planning." pith.science (2026). https://pith.science/paper/XHGOEB2D

@misc{pith2026190801362,
  author       = {Pith},
  title        = {Pith review of: ASNets: Deep Learning for Generalised Planning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/XHGOEB2D}},
  note         = {Machine review of arXiv:1908.01362}
}
read the original abstract

In this paper, we discuss the learning of generalised policies for probabilistic and classical planning problems using Action Schema Networks (ASNets). The ASNet is a neural network architecture that exploits the relational structure of (P)PDDL planning problems to learn a common set of weights that can be applied to any problem in a domain. By mimicking the actions chosen by a traditional, non-learning planner on a handful of small problems in a domain, ASNets are able to learn a generalised reactive policy that can quickly solve much larger instances from the domain. This work extends the ASNet architecture to make it more expressive, while still remaining invariant to a range of symmetries that exist in PPDDL problems. We also present a thorough experimental evaluation of ASNets, including a comparison with heuristic search planners on seven probabilistic and deterministic domains, an extended evaluation on over 18,000 Blocksworld instances, and an ablation study. Finally, we show that sparsity-inducing regularisation can produce ASNets that are compact enough for humans to understand, yielding insights into how the structure of ASNets allows them to generalise across a domain.

Figures

Figures reproduced from arXiv: 1908.01362 by the authors.

Figure 1
Figure 1. Part of the PPDDL description of a simple problem that we will use to illustrate the structure of ASNets. This is a toy navigation domain where a robot, shakey, is tasked with moving from place to place in a building using movement actions of the form drive(shakey, ?from, ?to). When invoked, a drive action moves the robot from its initial position to its destination successfully with 90% probability, and does nothin… view at source ↗
Figure 2
Figure 2. High-level overview of an ASNet. Each coloured rectangle represents an action module (red) or a proposition module (blue); these modules apply learnt trans￾formations to input feature vectors in order to produce more expressive output feature vectors. Information flows from the input (left) to the output (right) along the black lines connecting modules in successive layers. For the sake of visual clar￾ity, skip conn… view at source ↗
Figure 3
Figure 3. Schematic of an action module for the unreliable-robot problem ( [PITH_FULL_IMAGE:figures/full_fig_p009_3.png] view at source ↗
Figures from the paper (10 more)
Figure 4
Figure 4. Figure 4: A proposition module for the unreliable-robot problem ( [PITH_FULL_IMAGE:figures/full_fig_p011_4.png]
Figure 5
Figure 5. Figure 5: Illustration of the tire placement pattern in the first three Triangle Tireworld problems, adapted from Little and Thiébaux (2007). Black locations, including those around the outside edge, have spare tires; white locations do not. The road structure of larger instance…
Figure 6
Figure 6. Figure 6: Cumulative success rate of rollouts across a set of test problems for each probabilis￾tic domain. All ASNet configurations except the configuration without heuristic input (“no h.”) asymptotically achieve the same (perfect) coverage in CosaNostra, Triangle Tireworld, a…
Figure 7
Figure 7. Figure 7: Cumulative number of problems solved over time for deterministic test problems. Refer to main text for description of plotted quantities. LAMA-2011 is omitted: it achieved the same coverage as LAMA-first on our evaluation problems, but took much more time to execute on…
Figure 8
Figure 8. Figure 8: An easily-readable depiction of the sparse ASNet trained for the Triangle Tire￾world problem, along with comments describing the intuitive meaning of action and proposition layers (demarcated by boxes). The ELU activation functions have been omitted because all activat…
Figure 9
Figure 9. Figure 9: Visualisation of the activations of an ASNet for the first four actions in a successful rollout on Triangle Tireworld. The diagram for “step t” represents the action chosen by the ASNet and executed at time t. over zero or one corresponding φ 1 changetire(?loc) modules…
Figure 10
Figure 10. Figure 10: PPDDL domain for Triangle Tireworld. CosaNostra Pizza (Toyer et al., 2018) This domain was described in the main text of the article. A PPDDL description of the domain is given in [PITH_FULL_IMAGE:figures/full_fig_p058_10.png]
Figure 11
Figure 11. Figure 11: PPDDL domain for CosaNostra Pizza. grippers. Like Gold Miner, this domain first appeared in the learning track of the 2008 International Planning Competition. 59 [PITH_FULL_IMAGE:figures/full_fig_p059_11.png]
Figure 12
Figure 12. Figure 12: An easily-readable representation of a sparse ASNet trained for the CosaNostra Pizza domain. Weights have been rounded to two decimal places, and f(·) has been used to denote the ELU activation function. The middle column shows abbreviations used to refer to activatio…
Figure 13
Figure 13. Figure 13: Activations for the sparse CosaNostra Pizza policy given in [PITH_FULL_IMAGE:figures/full_fig_p062_13.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

85 extracted references · 76 canonical work pages

  1. [1]

    \ \ Fukunaga, A

    Asai, M. \ \ Fukunaga, A. 2018 . Classical planning in deep latent space: Bridging the subsymbolic-symbolic boundary \ In AAAI Conference on Artificial Intelligence (AAAI)

    2018

  2. [2]

    N., Garg, S., \ Mausam 2018

    Bajpai, A. N., Garg, S., \ Mausam 2018 . Transfer of deep reactive policies for mdp planning \ In Conference on Neural Information Processing Systems (NeurIPS)

    2018

  3. [3]

    G., Bradtke, S

    Barto, A. G., Bradtke, S. J., \ Singh, S. P. 1995 . Learning to act using real-time dynamic programming \ Artificial Intelligence

    1995

  4. [4]

    \ \ Tsitsiklis, J

    Bertsekas, D. \ \ Tsitsiklis, J. N. 1996 . Neuro-Dynamic Programming . Athena Scientific

    1996

  5. [5]

    \ \ de Raedt, L

    Blockeel, H. \ \ de Raedt, L. 1998 . Top-down induction of first-order logical decision trees \ Artificial Intelligence

    1998

  6. [6]

    Bonet, B., Frances, G., \ Geffner, H. 2019 . Learning features and abstract actions for computing generalized plans \ In AAAI Conference on Artificial Intelligence (AAAI) , 33, \ 2703--2710

    work page 2019

  7. [7]

    \ \ Geffner, H

    Bonet, B. \ \ Geffner, H. 2003 . Labeled RTDP : Improving the convergence of real-time dynamic programming \ In International Conference on Automated Planning and Scheduling (ICAPS)

    work page 2003

  8. [8]

    Features, Projections, and Representation Change for Generalized Planning

    Bonet, B. \ \ Geffner, H. 2018 . Features, projections, and representation change for generalized planning \ arXiv:1801.10055

    work page Pith review arXiv 2018

Show all 85 references
  1. [9]

    M., Bruna, J., LeCun, Y., Szlam, A., \ Vandergheynst, P

    Bronstein, M. M., Bruna, J., LeCun, Y., Szlam, A., \ Vandergheynst, P. 2017 . Geometric deep learning: going beyond Euclidean data \ IEEE Signal Processing Magazine

    2017

  2. [10]

    \ \ Buffet, O

    Bryce, D. \ \ Buffet, O. 2008 . 6th I nternational P lanning C ompetition: Uncertainty part \ In Workshop on the International Planning Competition

    2008

  3. [11]

    \ \ Aberdeen, D

    Buffet, O. \ \ Aberdeen, D. 2009 . The factored policy-gradient planner \ Artificial Intelligence

    2009

  4. [12]

    Camacho, E. F. \ \ Alba, C. B. 2007 . Model predictive control . Springer Science & Business Media

    2007

  5. [13]

    G., Jim \'e nez, S., L \'o pez, C

    Coles, A., Coles, A., Olaya, A. G., Jim \'e nez, S., L \'o pez, C. L., Sanner, S., \ Yoon, S. 2012 . A survey of the seventh international planning competition \ AI Magazine

    2012

  6. [14]

    Culberson, J. 1997 . S okoban is PSPACE -complete \ https://webdocs.cs.ualberta.ca/ joe/Preprints/Sokoban/paper.html

    1997

  7. [15]

    \ \ Fuentetaja, R

    de la Rosa, T. \ \ Fuentetaja, R. 2017 . Bagging strategies for learning planning policies \ Annals of Mathematics and Artificial Intelligence

    2017

  8. [16]

    de la Rosa, T., Jim \'e nez, S., Fuentetaja, R., \ Borrajo, D. 2011 . Scaling up heuristic planning with relational decision trees \ Journal of Artificial Intelligence Research (JAIR)

    2011

  9. [17]

    Dietterich, T. G. 2000 . Ensemble methods in machine learning \ Multiple classifier systems

    2000

  10. [18]

    K., Maclaurin, D., Iparraguirre, J., Bombarell, R., Hirzel, T., Aspuru-Guzik, A., \ Adams, R

    Duvenaud, D. K., Maclaurin, D., Iparraguirre, J., Bombarell, R., Hirzel, T., Aspuru-Guzik, A., \ Adams, R. P. 2015 . Convolutional networks on graphs for learning molecular fingerprints \ In Conference on Neural Information Processing Systems (NeurIPS)

    2015

  11. [19]

    Ferber, P., Helmert, M., \ Hoffman, J. 2020 . Neural network heuristics for classical planning: A study of hyperparameter space \ In ECAI

    2020

  12. [20]

    Fern, A., Khardon, R., \ Tadepalli, P. 2011 . The first learning track of the international planning competition \ Machine Learning

    2011

  13. [21]

    Fern, A., Yoon, S., \ Givan, R. 2004 . Approximate policy iteration with a policy language bias \ In Conference on Neural Information Processing Systems (NeurIPS)

    2004

  14. [22]

    B., Geissmann, C., \ Pommerening, F

    Franc \`e s, G., Corr \^e a, A. B., Geissmann, C., \ Pommerening, F. 2019 . Generalized potential heuristics for classical planning \ In International Joint Conferences on Artificial Intelligence (IJCAI)

    2019

  15. [23]

    Size-independent neural transfer for rddl planning \ In International Conference on Automated Planning and Scheduling (ICAPS)

    Garg, S., Bajpai, A., \ Mausam 2019 . Size-independent neural transfer for rddl planning \ In International Conference on Automated Planning and Scheduling (ICAPS)

    2019

  16. [24]

    Geffner, H. 2018a . International Joint Conferences on Artificial Intelligence (IJCAI)-ECAI keynote: Model-free, model-based, and general intelligence . Recording at https://www.youtube.com/watch?v=g3lc8BxTPiU&t=1906

    1906

  17. [25]

    Geffner, H. 2018b . Model-free, model-based, and general intelligence \ arXiv:1806.02308

    arXiv

  18. [26]

    Gomoluch, P., Alrajeh, D., \ Russo, A. 2019 . Learning classical planning strategies with policy gradient \ In International Conference on Automated Planning and Scheduling (ICAPS)

    2019

  19. [27]

    \ \ Thi \'e baux, S

    Gretton, C. \ \ Thi \'e baux, S. 2004 . Exploiting first-order regression in inductive policy selection \ In Conference on Uncertainty in Artificial Intelligence (UAI)

    2004

  20. [28]

    Groshev, E., Goldstein, M., Tamar, A., Srivastava, S., \ Abbeel, P. 2018 . Learning generalized reactive policies using deep neural networks \ In International Conference on Automated Planning and Scheduling (ICAPS)

    2018

  21. [29]

    \ \ Geffner, H

    Haslum, P. \ \ Geffner, H. 2000 . Admissible heuristics for optimal planning \ In International Conference on Artificial Intelligence Planning and Scheduling (AIPS)

    2000

  22. [30]

    Helmert, M. 2006 . The F ast D ownward planning system \ Journal of Artificial Intelligence Research (JAIR)

    2006

  23. [31]

    \ \ Domshlak, C

    Helmert, M. \ \ Domshlak, C. 2009 . Landmarks, critical paths and abstractions: what's the difference anyway? \ In International Conference on Automated Planning and Scheduling (ICAPS)

    2009

  24. [32]

    \ \ Schmidhuber, J

    Hochreiter, S. \ \ Schmidhuber, J. 1997 . Long short-term memory \ Neural computation

    1997

  25. [33]

    Hoffmann, J. 2001 . FF : The Fast-Forward planning system \ AI Magazine

    2001

  26. [34]

    \ \ De Giacomo, G

    Hu, Y. \ \ De Giacomo, G. 2011 . Generalized planning: Synthesizing plans that work for multiple environments \ In International Joint Conferences on Artificial Intelligence (IJCAI)

    2011

  27. [35]

    Issakkimuthu, M., Fern, A., \ Tadepalli, P. 2018 . Training deep reactive policies for probabilistic planning problems \ In International Conference on Automated Planning and Scheduling (ICAPS)

    2018

  28. [36]

    R., Savarese, S., \ Saxena, A

    Jain, A., Zamir, A. R., Savarese, S., \ Saxena, A. 2016 . Structural- RNN : Deep learning on spatio-temporal graphs \ In Computer Vision and Pattern Recognition (CVPR)

    2016

  29. [37]

    Jim \'e nez, S., de la Rosa, T., Fern \'a ndez, S., Fern \'a ndez, F., \ Borrajo, D. 2012 . A review of machine learning for automated planning \ The Knowledge Engineering Review

    2012

  30. [38]

    A., Eldawy, M., L \'a zaro-Gredilla, M., Lou, X., Dorfman, N., Sidor, S., Phoenix, S., \ George, D

    Kansky, K., Silver, T., M \'e ly, D. A., Eldawy, M., L \'a zaro-Gredilla, M., Lou, X., Dorfman, N., Sidor, S., Phoenix, S., \ George, D. 2017 . Schema networks: Zero-shot transfer with a generative causal model of intuitive physics \ In International Conference on Machine Lear...

    2017

  31. [39]

    L., Julian, K., \ Kochenderfer, M

    Katz, G., Barrett, C., Dill, D. L., Julian, K., \ Kochenderfer, M. J. 2017 . Reluplex: An efficient SMT solver for verifying deep neural networks \ In International Conference on Computer-Aided Verification (CAV)

    2017

  32. [40]

    Kearnes, S., McCloskey, K., Berndl, M., Pande, V., \ Riley, P. 2016 . Molecular graph convolutions: moving beyond fingerprints \ Journal of Computer-Aided Molecular Design

    2016

  33. [41]

    Khardon, R. 1999 . Learning action strategies for planning domains \ Artificial Intelligence

    1999

  34. [42]

    Krizhevsky, A., Sutskever, I., \ Hinton, G. E. 2012 . Imagenet classification with deep convolutional neural networks \ In Conference on Neural Information Processing Systems (NeurIPS)

    2012

  35. [43]

    \ \ Bengio, Y

    LeCun, Y. \ \ Bengio, Y. 1995 . Convolutional networks for images, speech, and time series \ The handbook of brain theory and neural networks

    1995

  36. [44]

    LeCun, Y., Bengio, Y., \ Hinton, G. 2015 . Deep learning \ Nature

    2015

  37. [45]

    E., \ Stoica, I

    Liaw, R., Liang, E., Nishihara, R., Moritz, P., Gonzalez, J. E., \ Stoica, I. 2018 . T une: A research platform for distributed model selection and training \ arXiv:1807.05118

    2018 arXiv

  38. [46]

    Lipton, Z. C. 2016 . The mythos of model interpretability \ arXiv:1606.03490

    2016 arXiv

  39. [47]

    \ \ Thi \'e baux, S

    Little, I. \ \ Thi \'e baux, S. 2007 . Probabilistic planning vs. replanning \ In International Conference on Automated Planning and Scheduling (ICAPS) workshops

    2007

  40. [48]

    Long, J., Shelhamer, E., \ Darrell, T. 2015 . Fully convolutional networks for semantic segmentation \ In Computer Vision and Pattern Recognition (CVPR)

    2015

  41. [49]

    \ \ Geffner, H

    Martin, M. \ \ Geffner, H. 2000 . Learning generalized policies in planning using concept languages \ In International Conference on Principles of Knowledge Representation and Reasoning (KR)

    2000

  42. [50]

    Mausam \ \ Kolobov, A. 2012 . Planning with Markov Decision Processes . Morgan & Claypool

    2012

  43. [51]

    Mnih, V., Kavukcuoglu, K., Silver, D., Graves, A., Antonoglou, I., Wierstra, D., \ Riedmiller, M. 2013 . Playing Atari with deep reinforcement learning \ In NIPS workshops

    2013

  44. [52]

    C., \ Kova c evi \'c , J

    Niu, S., Chen, S., Guo, H., Targonski, C., Smith, M. C., \ Kova c evi \'c , J. 2017 . Generalized value iteration networks: Life beyond lattices \ arXiv:1706.02416

    2017 arXiv

  45. [53]

    Quinlan, J. R. 1986 . Induction of decision trees \ Machine learning

    1986

  46. [54]

    \ \ Warmuth, M

    Ratner, D. \ \ Warmuth, M. K. 1986 . Finding a shortest solution for the n n extension of the 15-puzzle is intractable. \ In AAAI Conference on Artificial Intelligence (AAAI)

    1986

  47. [55]

    Richter, S., Westphal, M., \ Helmert, M. 2011 . LAMA 2008 and 2011 \ In International Planning Competition , \ 117--124

    2011

  48. [56]

    Rivest, R. L. 1987 . Learning decision lists \ Machine learning

    1987

  49. [57]

    Ross, S., Gordon, G., \ Bagnell, D. 2011 . A reduction of imitation learning and structured prediction to no-regret online learning \ In AISTATS

    2011

  50. [58]

    Q., \ Sanner, S

    Say, B., Wu, G., Zhou, Y. Q., \ Sanner, S. 2017 . Nonlinear hybrid planning with deep net learned transition models and mixed-integer linear programming \ In International Joint Conferences on Artificial Intelligence (IJCAI)

    2017

  51. [59]

    Shen, W., Trevizan, F., \ Thi \'e baux, S. 2020 . Learning domain-independent planning heuristics with hypergraph networks \ International Conference on Automated Planning and Scheduling (ICAPS)

    2020

  52. [60]

    W., Toyer, S., Thi \' e baux, S., \ Xie, L

    Shen, W., Trevizan, F. W., Toyer, S., Thi \' e baux, S., \ Xie, L. 2019 . Guiding search with generalized policies for probabilistic planning \ In Symposium on Combinatorial Search (SOCS)

    2019

  53. [61]

    Sievers, S., Katz, M., Sohrabi, S., Samulowitz, H., \ Ferber, P. 2019 . Deep learning for cost-optimal planning: Task-dependent planner selection \ In AAAI Conference on Artificial Intelligence (AAAI)

    2019

  54. [62]

    J., Guez, A., Sifre, L., Van Den Driessche, G., Schrittwieser, J., Antonoglou, I., Panneershelvam, V., \ Lanctot, M

    Silver, D., Huang, A., Maddison, C. J., Guez, A., Sifre, L., Van Den Driessche, G., Schrittwieser, J., Antonoglou, I., Panneershelvam, V., \ Lanctot, M. 2016 . Mastering the game of Go with deep neural networks and tree search \ Nature

    2016

  55. [63]

    \ \ Thi \'e baux, S

    Slaney, J. \ \ Thi \'e baux, S. 2001 . Blocks world revisited \ Artificial Intelligence

    2001

  56. [64]

    Sourek, G., Aschenbrenner, V., Zelezny, F., Schockaert, S., \ Kuzelka, O. 2018 . Lifted relational neural networks: Efficient learning of latent relational structures \ Journal of Artificial Intelligence Research (JAIR)

    2018

  57. [65]

    Srivastava, S., Immerman, N., Zilberstein, S., \ Zhang, T. 2011 . Directed search for generalized plans using classical planners. \ In International Conference on Automated Planning and Scheduling (ICAPS)

    2011

  58. [66]

    Tamar, A., Wu, Y., Thomas, G., Levine, S., \ Abbeel, P. 2016 . Value iteration networks \ In Conference on Neural Information Processing Systems (NeurIPS)

    2016

  59. [67]

    Tibshirani, R. 1996 . Regression shrinkage and selection via the lasso \ J.\@ Royal Stat.\@ Soc.: Series B (Methodological)

    1996

  60. [68]

    Y., \ Tedrake, R

    Tjeng, V., Xiao, K. Y., \ Tedrake, R. 2019 . Evaluating robustness of neural networks with mixed integer programming \ In International Conference on Learning Representations (ICLR)

    2019

  61. [69]

    Toyer, S., Trevizan, F., Thi \'e baux, S., \ Xie, L. 2018 . Action schema networks: Generalised policies with deep learning \ In AAAI Conference on Artificial Intelligence (AAAI)

    2018

  62. [70]

    Trevizan, F., Thi \'e baux, S., \ Haslum, P. 2017 . Occupation measure heuristics for probabilistic planning \ In International Conference on Automated Planning and Scheduling (ICAPS)

    2017

  63. [71]

    Trevizan, F. W. \ \ Veloso, M. M. 2012 . Short-sighted stochastic shortest path problems \ In International Conference on Automated Planning and Scheduling (ICAPS)

    2012

  64. [72]

    \ \ Blockeel, H

    Uwents, W. \ \ Blockeel, H. 2005 . Classifying relational data with neural networks \ In International Conference on Inductive Logic Programming (ILP)

    2005

  65. [73]

    L., Roberts, M., \ Sanner, S

    Vallati, M., Chrpa, L., Grze \'s , M., McCluskey, T. L., Roberts, M., \ Sanner, S. 2015 . The 2014 International Planning Competition : Progress and trends \ AI Magazine

    2015

  66. [74]

    N., Kaiser, ., \ Polosukhin, I

    Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., Kaiser, ., \ Polosukhin, I. 2017 . Attention is all you need \ In Conference on Neural Information Processing Systems (NeurIPS)

    2017

  67. [75]

    Vinyals, O., Fortunato, M., \ Jaitly, N. 2015 . Pointer networks \ In Conference on Neural Information Processing Systems (NeurIPS)

    2015

  68. [76]

    Weston, J., Chopra, S., \ Bordes, A. 2014 . Memory networks \ arXiv:1410.3916

    2014 arXiv

  69. [77]

    V., Norouzi, M., Macherey, W., Krikun, M., Cao, Y., Gao, Q., \ Macherey, K

    Wu, Y., Schuster, M., Chen, Z., Le, Q. V., Norouzi, M., Macherey, W., Krikun, M., Cao, Y., Gao, Q., \ Macherey, K. 2016 . G oogle's neural machine translation system: Bridging the gap between human and machine translation \ arXiv:1609.08144

    2016 arXiv

  70. [78]

    Xu, Y., Fern, A., \ Yoon, S. W. 2007 . Discriminative learning of beam-search heuristics for planning \ In International Joint Conferences on Artificial Intelligence (IJCAI)

    2007

  71. [79]

    W., Fern, A., \ Givan, R

    Yoon, S. W., Fern, A., \ Givan, R. 2007 . Using learned policies in heuristic-search planning \ In International Joint Conferences on Artificial Intelligence (IJCAI)

    2007

  72. [80]

    Yoon, S., Fern, A., \ Givan, R. 2006 . Discrepancy search with reactive policies for planning \ In AAAI-06 Workshop on Learning for Search

    2006

  73. [81]

    Yoon, S., Fern, A., \ Givan, R. 2002 . Inductive policy selection for first-order MDPs \ In Conference on Uncertainty in Artificial Intelligence (UAI)

    2002

  74. [82]

    Younes, H. L. \ \ Littman, M. L. 2004 . PPDDL1.0: an extension to PDDL for expressing planning domains with probabilistic effects \ , CMU

    2004

  75. [83]

    L., Littman, M

    Younes, H. L., Littman, M. L., Weissman, D., \ Asmuth, J. 2005 . The first probabilistic track of the international planning competition \ Journal of Artificial Intelligence Research (JAIR)

    2005

  76. [84]

    Zambaldi, V., Raposo, D., Santoro, A., Bapst, V., Li, Y., Babuschkin, I., Tuyls, K., Reichert, D., Lillicrap, T., Lockhart, E., Shanahan, M., Langston, V., Pascanu, R., Botvinick, M., Vinyals, O., \ Battaglia, P. 2019 . Deep reinforcement learning with relational inductive bia...

    2019

  77. [85]

    write newline

    " write newline "" before.all 'output.state := FUNCTION fin.entry add.period write newline FUNCTION new.block output.state before.all = 'skip after.block 'output.state := if FUNCTION new.sentence output.state after.block = 'skip output.state before.all = 'skip after.sentence '...

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.