Pith. sign in

REVIEW 4 major objections 6 minor 39 references

Group Effect Enhanced Generative Adversarial Imitation Learning for Individual Travel Behavior Modeling under Incentives

T0 review · 4 major / 6 minor · reviewed 2026-08-04 · deepseek-v4-flash

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

Pith's one-line read A group-conditioned GAIL model predicts individual transit riders' responses to fare incentives with higher accuracy than GAIL, cGAIL, and AIRL, and stays accurate under sparse or partial demonstrations.

desk verdict The empirical comparison is real, but the method is missing its environment—without that, the GAIL claims don't hold up. read the letter →

arxiv 2509.06656 v1 pith:SDGMZQSY submitted 2025-09-08 cs.LG

Yuanyuan Wu , Zhenlin Qin , Leizhen Wang , Xiaolei Ma , Zhenliang Ma This is my paper · ORCID
classification cs.LG
keywords imitationlearninggenerativeadversarialtravelbehaviorresponsefareincentivessmartcarddataindividualpredictiongroupeffectsparsity
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

gcGAIL is a generative adversarial imitation learning model that models how individual public transport passengers change their travel timing in response to a fare discount. The paper's central claim is that conditioning the policy and discriminator on passenger group features—schedule flexibility, inconvenience of shifting, and trip distance, each cut into quartiles—lets the model learn an individual's response from other similar passengers' trajectories, compensating for sparse personal data. On Hong Kong MTR Octopus card data around the 2014 'early bird' discount, it reports the best accuracy and F1 among BC, AIRL, GAIL, and cGAIL, and shows smaller performance drops when demonstrations are spatially partial or when whole passenger groups are omitted. The practical point is that a mobility authority could use such a model to forecast who will shift to off-peak travel and when, and to time personalized incentives.

What carries the argument

The load-bearing object is the group-conditioned policy πθ(a|s, gc) trained against a group-conditioned discriminator D(s, a, gc), with the advantage computed by Generalized Advantage Estimation and the update performed by PPO. The condition gc is the new ingredient: three passenger attributes (schedule flexibility, incentive inconvenience, trip distance), binned into quartiles, that label each trajectory and allow the model to borrow statistical strength from other passengers in the same group. What it does is turn 'learn this one passenger's policy' into 'learn the policy shared by a group, then specialize through the condition,' which is the mechanism the paper credits for robustness to s

What would settle it

Re-run Experiment I after randomly permuting group labels across passengers while keeping each passenger's trajectory intact; if prediction accuracy does not fall to approximately cGAIL's level, the group-effect enrichment is not the mechanism driving the improvement.

Watch

Extended reading notes

Core claim

Stripped to its central claim, the paper says that conditioning an imitation-learning policy on coarse passenger-group labels lets one individual's travel response be learned from the group's shared patterns. The setting is a fare-discount program: each passenger's month is encoded as state features and a binary action (off-peak vs peak), and the gcGAIL actor and discriminator both receive a condition vector made from three quartile-coded attributes—schedule flexibility, inconvenience of shifting, and trip distance—computed from two months of pre-discount card data. The paper reports that this group-conditioned model reaches 0.95 accuracy and 0.91 F1, above cGAIL, GAIL, and AIRL, and that it

Load-bearing premise

The paper assumes that travelers' month-to-month decisions follow a Markov decision process with a deterministic time-series transition, but it never defines that transition or how training rollouts are generated; if rollouts are just replayed observed sequences, the model is a sequence classifier and the imitation-learning framing is not load-bearing.

Editorial extensions

If this is right

  • Operators could query the trained policy at any month to predict which passengers will shift to off-peak travel, enabling personalized incentive timing rather than a uniform discount.
  • Passengers with very few personal observations can still be modeled accurately once their three group attributes are computed from a short pre-policy observation window.
  • Training on a subset of stations or on 10% of riders still yields strong predictions, so data-collection campaigns for new incentive programs may need less coverage than previously assumed.
  • Excluding the most-responsive passenger groups from training degrades gcGAIL less than GAIL or cGAIL, meaning policies can be built even when the most informative demonstrations are missing.

Reading between the lines

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

  • Because the transition function is never specified, a fair-minded reader can treat the learned object as a sequential classifier; a useful test is whether a supervised sequence model with the same group conditioning reaches the same accuracy, which would separate the group-effect idea from the RL machinery.
  • The quartile coding of flexibility, inconvenience, and distance is a coarse discretization; replacing it with continuous embeddings could extract more signal from the same smart-card features, at some cost in interpretability.
  • If group effects transfer across networks, the same conditional architecture could be pretrained on one city's card data and fine-tuned on a smaller sample from another city—a transfer setting the paper does not test.
  • The adopter labels come from a change-point detector; the reported adopter accuracy therefore blends detection error with prediction error, so an end-to-end model that learns the change points jointly with the policy might give cleaner estimates of response timing.
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 / 6 minor

Summary. The paper proposes gcGAIL, a group-effect-conditioned variant of Generative Adversarial Imitation Learning, and applies it to model individual passengers' monthly travel-mode choices (off-peak vs. peak) in response to a real fare-discount program. The method augments a conditional GAIL policy with passenger-group labels derived from pre-promotion smart-card travel characteristics, trains via PPO with GAE and an adversarial discriminator (Eqs. 2-7, Algorithm 1), and is evaluated on Hong Kong MTR 'early-bird' data. Experiments compare gcGAIL with BC, AIRL, GAIL, and cGAIL, reporting higher accuracy and F1 in the full-data experiment (Table 3) and stable performance under reduced spatial coverage, smaller demonstration sets, and excluded passenger groups (Experiments II-IV). The central claim is that gcGAIL outperforms these methods in accuracy, generalization, and demonstration efficiency.

Significance. If the method is fully specified and reproducible, the contribution is a practically relevant, incremental extension of conditional GAIL: using group-level behavioral regularities to mitigate data sparsity in individual-level imitation learning from smart-card data. The paper's strengths are the use of a real-world AFC panel dataset, the structured ablation over conditional and group effects, and the explicit comparison against AIRL/cGAIL. However, the machine-learning contribution is modest, there are no theoretical results or released code, and the central empirical claim is weakened by an underspecified training environment. The reported numbers would support the headline comparison only after the missing environment definition and evaluation details are supplied; the paper is not yet reproducible in its current form.

major comments (4)
  1. [Section 3.1 / Algorithm 1] The environment and transition function f used for policy rollouts are never specified. The only statement is "The state transition is deterministic as time series" (Section 3.1), which does not define how s_{t+1} is computed from (s_t, a_t), whether a_t affects future states, or what initial-state distribution and horizon are used. Algorithm 1 line 3 requires sampling trajectories τπ in the environment, and Eqs. 5-6 rely on next-state values and rewards. Without this, the trained object is empirically a sequence classifier rather than a policy from an explicit MDP, and the reported GAIL/PPO results are not reproducible. The authors should specify f exactly (including how λ_t, λ_{t-1}, and ms_t evolve), describe the rollout procedure used in Stable-Baselines3, and discuss the degenerate case f(s'|s,a)=f(s'|s) if actions do not affect transitions.
  2. [Table 3, Section 3.4] The BC baseline is trivial: accuracy=0.28, recall=1.00, precision=0.27 means the classifier predicts the positive class almost always, which is equivalent to a no-information rule when the positive class has prevalence about 28%. This does not constitute a trained behavioral-cloning benchmark and misleads the comparison. The authors should implement a standard BC baseline (e.g., a supervised classifier on the same state/group features with proper class balancing or thresholding) and report its performance, or remove BC from the headline comparison.
  3. [Section 3.6 / Figure 7 / Table 2] The robustness claims are based on single-seed runs. Table 2 fixes random seed to 0, and the '±' values in Table 3 appear to be temporal dispersion across months rather than variability across repeated experiments. Since Figure 7 shows small differences and even an unexplained performance decline at 70% for GAIL/cGAIL, the conclusion that gcGAIL is robust to data sparsity and passenger diversity requires repeated runs with mean±std over seeds/splits and, where possible, statistical significance tests. Please also report the number of independent runs and the source of uncertainty for each reported metric.
  4. [Section 3.2 / Table 1 / Experiment IV] The group labels and the 'most responsive' groups used in Experiment IV are imported from prior papers (Wang et al., 2023; Wu et al., 2025) without self-contained definitions. The quartile features flex_u, con_u, dis_u are described in words, but the exact computation formulas and their empirical distributions are not given. Moreover, the scenarios wf3/wc2/wd2/ges exclude groups that were identified as 'more responsive' using the same dataset and outcome. This raises a potential circularity concern: the generalization claim may partly reflect the way the scenarios were constructed rather than a property of the model. Provide precise definitions, show that the group quartiles are computed only from pre-promotion data without outcome leakage, and ideally validate the generalization on a temporally or spatially held-out data split.
minor comments (6)
  1. [Abstract / Section 3] 'Pattern demonstration efficiency' is used in the abstract and conclusions but never defined. Please define the metric or rephrase.
  2. [Section 3.1] The sentence "The state transition is deterministic as time series, we need to inversely reveal the reward function and policy function" is grammatically incomplete and should be replaced by a formal transition-function specification.
  3. [Table 4] Table 4 reports only the 'ges' scenario. For complete reporting, include the full metrics for all Experiment IV scenarios (wf3, wc2, wd2, ges).
  4. [Section 3.2] The 80/20 random split is mentioned, but no validation split or hyperparameter-selection procedure is described. Please state how the hyperparameters in Table 2 were chosen.
  5. [Figure 7] The radar/polygon chart mixes scenario labels ('10-100', 'ges', 'wf3', etc.) on one axis; a clearer legend or separate panels would improve readability, especially in grayscale.
  6. [General] No code or data availability statement is provided. Releasing the implementation and, where permitted, processed data would substantially aid reproducibility.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the reported gains are empirical on held-out data; the self-citations and underspecified transition are not load-bearing circular reductions.

full rationale

The paper's central claim is an empirical comparison of gcGAIL against baseline methods on an 80/20 held-out split of Octopus-card panel data (Section 3.2), with accuracy and F1 reported in Tables 3–4 and Figures 5–8. No fitted parameter is renamed as a prediction, and no equation reduces to the input by construction. The group features (flexibility, inconvenience, travel distance) are computed from the two months before the promotion (Section 3.1: 'calculated based on each user’s Octopus card activity during the two months prior to the promotion policy implementation'), so they are exogenous to the target response period and do not encode the label. The authors' self-citations (Wang et al., 2023; Wu et al., 2025) are used to select group features and define adopter types, but the load-bearing evidence is the independent experimental comparison, not the citations. The only plausible circularity candidates are the state features λ_t and ms_t, which could be misread as contemporaneous outcomes; however, the paper describes them as 'historical behavior' and 'accumulated incentives' (Section 2.1), and the very low BC accuracy (0.28) is inconsistent with label-in-state leakage. The statement 'The state transition is deterministic as time series' (Section 3.1) is an underspecified environment, creating a reproducibility/correctness risk for Algorithm 1's rollouts, but it is not a circular reduction of a prediction to its input. Overall, no specific circular step can be exhibited.

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

The central claim depends on a small number of hand-chosen preprocessing choices (quartile grouping, binary mode threshold), a single data split, and domain assumptions imported from the authors' prior papers. No new physical or mathematical entities are introduced. The most fragile assumption is the unspecified MDP environment needed for GAIL rollouts.

free parameters (4)
  • Group quantization bins = 4 quartile bins per group feature
    Flexibility, inconvenience, and distance are discretized into quartiles to form conditioning labels. The choice of 4 bins and the selected features (from prior work) defines the group effect; a different quantization would change the conditioning information.
  • Monthly mode cutoff = 0.5 (off-peak trips > half peak trips)
    A passenger's monthly behavior is labeled off-peak (action 1) if off-peak trips exceed half their peak trips. This binary threshold defines the target action and therefore the accuracy metrics.
  • Data split and seed = 80/20 train/test split, random seed 0
    A single random split is used; headline metrics come from one seed, with no repeated-seed error bars. The unexplained accuracy decline at 70% demonstration size in Experiment III is attributed to randomness.
  • Model hyperparameters = lr=1e-4, gamma=0.95, lambda=0.95, clip=0.2, hidden=64, batch=256
    Standard PPO/GAIL hyperparameters chosen by hand. They affect training stability but are not the central scientific claim.
assumptions (4)
  • domain assumption Expert trajectories arise from an optimal policy under some unknown reward function
    Used in Section 2.2, Eq. (1) following Ng et al. (2000); the IRL premise underlying why imitation of expert data should yield the traveler's decision strategy.
  • domain assumption Flexibility, inconvenience, and travel distance capture the relevant between-passenger heterogeneity for incentive response
    These group features are selected from the authors' prior work (Wang et al., 2023; Wu et al., 2025) rather than derived in this paper; the claimed benefit of group conditioning rests on them.
  • ad hoc to paper The state transition can be treated as deterministic time series
    Stated in Section 3.1 ('The state transition is deterministic as time series'). No transition function, simulator, or rollout environment is provided, yet Algorithm 1 requires sampling trajectories with the current policy.
  • domain assumption Binary action space (off-peak vs. other) adequately represents travel behavior response
    The model collapses continuous departure-time choices into a binary monthly label, so all predictions and metrics concern this coarse outcome.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Group Effect Enhanced Generative Adversarial Imitation Learning for Individual Travel Behavior Modeling under Incentives." pith.science (2026). https://pith.science/paper/SDGMZQSY

@misc{pith2026250906656,
  author       = {Pith},
  title        = {Pith review of: Group Effect Enhanced Generative Adversarial Imitation Learning for Individual Travel Behavior Modeling under Incentives},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/SDGMZQSY}},
  note         = {Machine review of arXiv:2509.06656}
}
read the original abstract

Understanding and modeling individual travel behavior responses is crucial for urban mobility regulation and policy evaluation. The Markov decision process (MDP) provides a structured framework for dynamic travel behavior modeling at the individual level. However, solving an MDP in this context is highly data-intensive and faces challenges of data quantity, spatial-temporal coverage, and situational diversity. To address these, we propose a group-effect-enhanced generative adversarial imitation learning (gcGAIL) model that improves the individual behavior modeling efficiency by leveraging shared behavioral patterns among passenger groups. We validate the gcGAIL model using a public transport fare-discount case study and compare against state-of-the-art benchmarks, including adversarial inverse reinforcement learning (AIRL), baseline GAIL, and conditional GAIL. Experimental results demonstrate that gcGAIL outperforms these methods in learning individual travel behavior responses to incentives over time in terms of accuracy, generalization, and pattern demonstration efficiency. Notably, gcGAIL is robust to spatial variation, data sparsity, and behavioral diversity, maintaining strong performance even with partial expert demonstrations and underrepresented passenger groups. The gcGAIL model predicts the individual behavior response at any time, providing the basis for personalized incentives to induce sustainable behavior changes (better timing of incentive injections).

Figures

Figures reproduced from arXiv: 2509.06656 by the authors.

Figure 1
Figure 1. gcGAIL model for individual travel behavior learning. The policy network acts as the Generator [PITH_FULL_IMAGE:figures/full_fig_p007_1.png] view at source ↗
Figure 2
Figure 2. Network, critical links (red arrows), and eligible promotion stations (dashed area) (adopted from [PITH_FULL_IMAGE:figures/full_fig_p009_2.png] view at source ↗
Figure 3
Figure 3. Model prediction accuracy of different models over time of (a.) non-Adopters and (b.) Adopters. [PITH_FULL_IMAGE:figures/full_fig_p013_3.png] view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: Model prediction accuracy of GAIL-related models over time upon (a.) non-Adopters and (b.) [PITH_FULL_IMAGE:figures/full_fig_p014_4.png]
Figure 5
Figure 5. Figure 5: Model prediction accuracy of GAIL, cGAIL, gcGAIL upon stations in the first and second ex [PITH_FULL_IMAGE:figures/full_fig_p015_5.png]
Figure 6
Figure 6. Figure 6: Model prediction accuracy of GAIL, cGAIL, gcGAIL over time of (a.) non-Adopters and (b.) [PITH_FULL_IMAGE:figures/full_fig_p016_6.png]
Figure 7
Figure 7. Figure 7: Model prediction accuracy and F1-Score of GAIL, cGAIL, gcGAIL from different scenarios. [PITH_FULL_IMAGE:figures/full_fig_p017_7.png]
Figure 8
Figure 8. Figure 8: Model prediction accuracy on different types of adopters. Results are from Experiment I. [PITH_FULL_IMAGE:figures/full_fig_p018_8.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

39 extracted references · 33 canonical work pages

  1. [1]

    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 ":" * " " * FUNCTION f...

  2. [2]

    , author Graham, D.J

    author Anupriya, A. , author Graham, D.J. , author H \"o rcher, D. , author Anderson, R. , year 2018 . title The impact of early bird scheme on commuter trip scheduling in H ong K ong: A causal analysis using travel card data . journal Transportation Research Board 97th Annual Meeting

    work page 2018

  3. [3]

    , author McFadden, D

    author Ben-Akiva, M. , author McFadden, D. , author Train, K. , author Walker, J. , author Bhat, C. , author Bierlaire, M. , author Bolduc, D. , author Boersch-Supan, A. , author Brownstone, D. , author Bunch, D.S. , et al., year 2002 . title Hybrid choice models: Progress and challenges . journal Marketing Letters volume 13 , pages 163--175

    work page 2002

  4. [4]

    , author Lerman, S.R

    author Ben-Akiva, M.E. , author Lerman, S.R. , year 1985 . title Discrete choice analysis: theory and application to travel demand . volume volume 9 . publisher MIT press

    work page 1985

  5. [5]

    , author Wulfe, B

    author Bhattacharyya, R. , author Wulfe, B. , author Phillips, D.J. , author Kuefler, A. , author Morton, J. , author Senanayake, R. , author Kochenderfer, M.J. , year 2022 . title Modeling human driving behavior through generative adversarial imitation learning . journal IEEE Transactions on Intelligent Transportation Systems volume 24 , pages 2874--2887

    work page 2022

  6. [6]

    , author Kim, J

    author Choi, S. , author Kim, J. , author Yeo, H. , year 2021 . title Trajgail: Generating urban vehicle trajectories using generative adversarial imitation learning . journal Transportation Research Part C: Emerging Technologies volume 128 , pages 103091

    work page 2021

  7. [7]

    , author Santana, E

    author Codevilla, F. , author Santana, E. , author L \'o pez, A.M. , author Gaidon, A. , year 2019 . title Exploring the limitations of behavior cloning for autonomous driving , in: booktitle Proceedings of the IEEE/CVF international conference on computer vision , pp. pages 9329--9338

    work page 2019

  8. [8]

    , author He, Q

    author Cui, Y. , author He, Q. , author Khani, A. , year 2018 . title Travel behavior classification: an approach with social network and deep learning . journal Transportation research record volume 2672 , pages 68--80

    work page 2018

Show all 39 references
  1. [9]

    , author Li, Y

    author Feng, J. , author Li, Y. , author Zhang, C. , author Sun, F. , author Meng, F. , author Guo, A. , author Jin, D. , year 2018 . title Deepmove: P redicting human mobility with attentional recurrent networks , in: booktitle Proceedings of the 2018 world wide web conferenc...

    2018

  2. [10]

    , author Luo, K

    author Fu, J. , author Luo, K. , author Levine, S. , year 2017 . title Learning robust rewards with adversarial inverse reinforcement learning . journal arXiv preprint arXiv:1710.11248

    2017 arXiv

  3. [11]

    , author Pouget-Abadie, J

    author Goodfellow, I. , author Pouget-Abadie, J. , author Mirza, M. , author Xu, B. , author Warde-Farley, D. , author Ozair, S. , author Courville, A. , author Bengio, Y. , year 2014 . title Generative adversarial nets . journal Advances in neural information processing syste...

    2014

  4. [12]

    , author Koutsopoulos, H.N

    author Halvorsen, A. , author Koutsopoulos, H.N. , author Ma, Z. , author Zhao, J. , year 2020 . title Demand management of congested public transport systems: A conceptual framework and application using smart card data . journal Transportation volume 47 , pages 2337--2365

    2020

  5. [13]

    , author Douglas, N

    author Henn, L. , author Douglas, N. , author Sloan, K. , year 2011 . title Surveying S ydney rail commuters’ willingness to change travel time , in: booktitle 34th Australasian Transport Research Forum , address Adelaide, Australia . p. pages 0101

    2011

  6. [14]

    , author Ermon, S

    author Ho, J. , author Ermon, S. , year 2016 . title Generative adversarial imitation learning . journal Advances in neural information processing systems volume 29

    2016

  7. [15]

    , author Kim, D.K

    author Kim, E.J. , author Kim, D.K. , author Sohn, K. , year 2022 . title Imputing qualitative attributes for trip chains extracted from smart card data using a conditional generative adversarial network . journal Transportation Research Part C: Emerging Technologies volume 13...

    2022

  8. [16]

    , author Morton, J

    author Kuefler, A. , author Morton, J. , author Wheeler, T. , author Kochenderfer, M. , year 2017 . title Imitating driver behavior with generative adversarial networks , in: booktitle 2017 IEEE intelligent vehicles symposium (IV) , organization IEEE . pp. pages 204--211

    2017

  9. [17]

    , author Jiang, H

    author Liu, S. , author Jiang, H. , year 2022 . title Personalized route recommendation for ride-hailing with deep inverse reinforcement learning and real-time traffic conditions . journal Transportation Research Part E: Logistics and Transportation Review volume 164 , pages 102780

    2022

  10. [18]

    , author Jiang, H

    author Liu, S. , author Jiang, H. , author Chen, S. , author Ye, J. , author He, R. , author Sun, Z. , year 2020 . title Integrating dijkstra’s algorithm into deep inverse reinforcement learning for food delivery route planning . journal Transportation Research Part E: Logisti...

    2020

  11. [19]

    , author Koutsopoulos, H.N

    author Ma, Z. , author Koutsopoulos, H.N. , author Liu, T. , author Basu, A.A. , year 2020 . title Behavioral response to promotion-based public transport demand management: L ongitudinal analysis and implications for optimal promotion design . journal Transportation Research ...

    2020

  12. [20]

    , author Osindero, S

    author Mirza, M. , author Osindero, S. , year 2014 . title Conditional generative adversarial nets . journal arXiv preprint arXiv:1411.1784

    2014 arXiv

  13. [21]

    title Early bird discount

    author MTRHongKong , year 2014 . title Early bird discount . https://www.mtr.com.hk/en/customer/main/early_bird.html

    2014

  14. [22]

    , author Russell, S

    author Ng, A.Y. , author Russell, S. , et al., year 2000 . title Algorithms for inverse reinforcement learning. , in: booktitle Icml , p. pages 2

    2000

  15. [23]

    , author Pajarinen, J

    author Osa, T. , author Pajarinen, J. , author Neumann, G. , author Bagnell, J.A. , author Abbeel, P. , author Peters, J. , et al., year 2018 . title An algorithmic perspective on imitation learning . journal Foundations and Trends in Robotics volume 7 , pages 1--179

    2018

  16. [24]

    , author Huang, W

    author Pan, M. , author Huang, W. , author Li, Y. , author Zhou, X. , author Luo, J. , year 2020 . title xgail: Explainable generative adversarial imitation learning for explainable human decision analysis , in: booktitle Proceedings of the 26th ACM SIGKDD International Confer...

    2020

  17. [25]

    , author Zhang, P

    author Qin, Z. , author Zhang, P. , author Ma, Z. , year 2024 . title Deepags: Deep learning with activity, geography and sequential information in predicting an individual's next trip destination . journal IET Intelligent Transport Systems volume 18 , pages 1895--1909

    2024

  18. [26]

    , author Hill, A

    author Raffin, A. , author Hill, A. , author Gleave, A. , author Kanervisto, A. , author Ernestus, M. , author Dormann, N. , year 2021 . title Stable-baselines3: Reliable reinforcement learning implementations . journal Journal of Machine Learning Research volume 22 , pages 1-...

    2021

  19. [27]

    , author Wolski, F

    author Schulman, J. , author Wolski, F. , author Dhariwal, P. , author Radford, A. , author Klimov, O. , year 2017 . title Proximal policy optimization algorithms . journal arXiv preprint arXiv:1707.06347

    2017 arXiv

  20. [28]

    , author Li, D

    author Song, Y. , author Li, D. , author Ma, Z. , author Liu, D. , author Zhang, T. , year 2024 . title A state-based inverse reinforcement learning approach to model activity-travel choices behavior with reward function recovery . journal Transportation Research Part C: Emerg...

    2024

  21. [29]

    , author Chen, X

    author Sun, L. , author Chen, X. , author He, Z. , author Miranda-Moreno, L.F. , year 2023 . title Routine pattern discovery and anomaly detection in individual travel behavior . journal Networks and Spatial Economics volume 23 , pages 407--428

    2023

  22. [30]

    , author Barto, A.G

    author Sutton, R.S. , author Barto, A.G. , et al., year 1998 . title Reinforcement learning: An introduction . volume volume 1 . publisher MIT press Cambridge

    1998

  23. [31]

    , author Warnell, G

    author Torabi, F. , author Warnell, G. , author Stone, P. , year 2018 . title Behavioral cloning from observation . journal arXiv preprint arXiv:1805.01954

    2018 arXiv

  24. [32]

    , author Chen, X

    author Wang, L. , author Chen, X. , author Ma, Z. , author Zhang, P. , author Mo, B. , author Duan, P. , year 2023 . title Data-driven analysis and modeling of individual longitudinal behavior response to fare incentives in public transport . journal Transportation , pages 1--24

    2023

  25. [33]

    , author Mo, B

    author Wang, S. , author Mo, B. , author Zheng, Y. , author Hess, S. , author Zhao, J. , year 2021 . title Comparing hundreds of machine learning classifiers and discrete choice models in predicting travel behavior: an empirical benchmark . journal arXiv preprint arXiv:2102.01130

    2021 arXiv

  26. [34]

    , author Markham, A

    author Wu, Y. , author Markham, A. , author Wang, L. , author Solus, L. , author Ma, Z. , year 2025 . title Data-driven causal behaviour modelling from trajectory data: A case for fare incentives in public transport . journal Journal of Public Transportation volume 27 , pages 100114

    2025

  27. [35]

    , author Li, Y

    author Zhang, X. , author Li, Y. , author Zhou, X. , author Luo, J. , year 2019 . title Unveiling taxi drivers' strategies via cgail: Conditional generative adversarial imitation learning , in: booktitle 2019 IEEE international conference on data mining (ICDM) , organization I...

    2019

  28. [36]

    , author Li, Y

    author Zhang, X. , author Li, Y. , author Zhou, X. , author Luo, J. , year 2020 . title cgail: Conditional generative adversarial imitation learning—an application in taxi drivers’ strategy learning . journal IEEE transactions on big data volume 8 , pages 1288--1300

    2020

  29. [37]

    , author Koutsopoulos, H.N

    author Zhao, Z. , author Koutsopoulos, H.N. , author Zhao, J. , year 2018 . title Detecting pattern changes in individual travel behavior: A bayesian approach . journal Transportation research part B: methodological volume 112 , pages 73--88

    2018

  30. [38]

    , author Liang, Y

    author Zhao, Z. , author Liang, Y. , year 2023 . title A deep inverse reinforcement learning approach to route choice modeling with context-dependent rewards . journal Transportation Research Part C: Emerging Technologies volume 149 , pages 104079

    2023

  31. [39]

    , author Maas, A.L

    author Ziebart, B.D. , author Maas, A.L. , author Bagnell, J.A. , author Dey, A.K. , et al., year 2008 . title Maximum entropy inverse reinforcement learning. , in: booktitle Aaai , organization Chicago, IL, USA . pp. pages 1433--1438

    2008

Pith tools

Reviewed August 4, 2026 · model on record in the stance chip above.