Pith. sign in

REVIEW 4 major objections 7 minor 25 references

Temporal-IRL: Modeling Port Congestion and Berth Scheduling with Inverse Reinforcement Learning

T0 review · 4 major / 7 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read Temporal-IRL learns a port's tacit berth-scheduling rules from AIS ship-tracking data and uses the learned reward to predict vessel actions, congestion, and departure windows.

desk verdict First IRL application to berth scheduling on real port data, but the headline accuracy is not yet credible without trivial baselines and error bars. read the letter →

arxiv 2506.19843 v1 pith:WAINY5TW submitted 2025-06-24 cs.AI

Guo Li , Zixiang Xu , Wei Zhang , Yikuan Hu , Xinyu Yang , Nikolay Aristov , Mingjie Tang , Elenna R Dugundji This is my paper · ORCID
classification cs.AI
keywords inversereinforcementlearningberthschedulingportcongestionpredictionLSTMautoencodertemporalfeatureextractionAISdatamaximumentropyIRLvesseldwelltime
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

This paper proposes Temporal-IRL, a method that treats a container terminal's berth scheduling as an inverse problem: instead of predicting congestion directly, it first recovers the reward function that explains the port's historical assignment decisions and then uses that reward to predict what happens next. The model rebuilds eight years of vessel movements at Maher Terminal into expert trajectories over 8-hour windows, represents each ship by size, carrier, and time already in port, and augments these features with temporal encoding from an LSTM autoencoder. Under maximum entropy inverse reinforcement learning, the inferred reward yields the next scheduling action for every vessel, a congestion flag (three or more vessels waiting), and the window in which a vessel leaves. On held-out data the paper reports 82.64% action accuracy, 74.06% congestion accuracy, and a claimed best-in-class 71.51% departure-window accuracy, beating LSTM, FNN, FNN-IRL, and XGBoost on most metrics. A reader should care because ports keep their scheduling logic tacit, and a learned reward makes that logic usable for supply-chain planning.

What carries the argument

The load-bearing mechanism is the maximum entropy inverse reinforcement learning loop, which models the port's scheduling policy as $\pi^*(a|s)=\exp(Q(s,a))/Z(s)$ with $Q$ defined by the Bellman expectation equation and $Z(s)$ the normalizing partition function; the trainable object is the reward $R(s,a)$. Before IRL sees a state, an LSTM autoencoder turns the 19-slot state vector — each slot an IMO expanded to (size, carrier, stay time) — into compressed temporal features, which the paper finds is the decisive component: replacing it with a feedforward network drops action accuracy from 82.64% to 70.01%. The IRL loop then uses the log-likelihood of observed expert state-action pairs to fit the reward, and the resulting probability distribution over actions is what yields the congestion and departure-window predictions.

What would settle it

Compare a variant of Temporal-IRL that adds origin, customs status, berth size, channel depth, and a contract-priority index as reward features on the same Maher test split; if leave accuracy rises materially above 71.51% or the learned reward assigns substantial weight to these added features, the three-feature model has not captured the port's true scheduling logic.

Watch

Extended reading notes

Core claim

On its own terms, the paper's discovery is that the port's berth-allocation behavior can be reconstructed from AIS positions alone and then explained by a reward function learned through maximum entropy IRL, without any direct communication from port operators. The terminal is modeled as 19 ordered slots — six berths, seven waiting positions, six incoming vessels — and each vessel is described by three attributes: size, carrier, and elapsed stay time. A long short-term memory autoencoder compresses the temporal sequence of these attributes, and the compressed features feed a MaxEnt IRL loop whose Boltzmann policy assigns probabilities to ten possible actions per slot: do nothing, stay, move to waiting, move to one of six berths, or leave the system. From that policy the model predicts the next scheduling action, declares congestion when three or more vessels are waiting, and outputs the departure time window. Tested on Maher Terminal data from January 2015 to September 2023, the authors report 82.64% action accuracy, 74.06% congestion accuracy, and 71.51% departure-window accuracy, the last of which they state is the best among the compared models (LSTM, FNN, FNN-IRL, and XGBoost).

Load-bearing premise

The argument collapses if the three features the model keeps — vessel size, carrier, and time already spent in port — do not carry the information that actually drives berth assignments, because the paper itself lists origin, customs status, berth size, channel depth, and contractual agreements as scheduling factors that the model never sees.

Editorial extensions

If this is right

  • Port scheduling logic can be extracted from ship-tracking data alone, so terminals that do not publish their berth rules can still be modeled for forecasting.
  • Congestion warnings can be issued one 8-hour step ahead with roughly 74% accuracy (XGBoost is slightly better at 75.89%), which supports logistics planning even when absolute precision is lacking.
  • Departure-window forecasts at 71.51%, the best among the compared models, translate directly into berth-release and inventory-timing estimates for downstream supply chain decisions.
  • The large gap between Temporal-IRL's 82.64% and FNN-IRL's 70.01% action accuracy shows that temporal feature extraction is responsible for most of the gain over non-temporal IRL.
  • Because the model emits a full scheduling action for every slot, it can be used for simulation: replaying the learned policy under hypothetical arrivals produces synthetic congestion scenarios.

Reading between the lines

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

  • The learned reward itself is a research artifact the paper does not fully exploit: reading which slot-action pairs carry high reward could reveal which vessel attributes the port actually prioritizes, turning the model into a measurement tool.
  • Adding the factors listed in Section 3.1.1 but omitted from the features (origin, customs, berth size, channel depth, contracts) is a natural extension; if accuracy improves, the current three-feature reward is a deliberate simplification, and if it does not, the omitted factors are not decision-relevant at this terminal.
  • The method is presented as terminal-agnostic, but only one terminal is tested; a multi-terminal or post-2023 evaluation would reveal whether the reward function captures general scheduling logic or just Maher-specific patterns.
  • The binary congestion definition (three or more vessels waiting) is an editorial choice grounded in the data; a graded congestion scale might be better for practical use, and the same IRL reward could be evaluated against it.
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

4 major / 7 minor

Summary. The paper proposes Temporal-IRL, a pipeline that combines LSTM-AE feature extraction with maximum-entropy inverse reinforcement learning to model berth scheduling at the Maher Terminal in the Port of New York/New Jersey. The port is discretized into 19 slots (6 berths, 7 waiting-area positions, 6 incoming vessels), and scheduling decisions are represented as per-slot one-hot actions over 10 action types. Using AIS-based data from January 2015 to September 2023, the authors train the model on historical state-action trajectories, infer a reward function, and use it to predict next-step actions, congestion states, and vessel departure time windows. They report action accuracy of 82.64%, congestion accuracy of 74.06%, and leave accuracy of 71.51%, comparing against LSTM, FNN, FNN-IRL, and XGBoost baselines on the same data.

Significance. If the reported results are robust, the paper would demonstrate a useful application of IRL to a practical logistics problem, with the temporal feature-extraction component as a plausible contribution. The temporal train/test split is a reasonable evaluation choice, and the inclusion of several baselines is commendable. However, the central quantitative claims need additional support: the action-accuracy metric is vulnerable to class-imbalance artifacts, no uncertainty quantification is provided, and the reported state features omit several factors the paper itself identifies as influential. The work is a promising case study, but the evidence as presented is insufficient to establish that Temporal-IRL learns port scheduling priorities better than trivial baselines.

major comments (4)
  1. [Section 4.2.2, Eq. (7), Table 4] The action-accuracy metric is computed element-wise over a 190-dimensional one-hot vector (19 slots times 10 actions), with empty slots encoded as an all-zero 'Nothing' vector. Section 3.2.2 and Figure 3 state that 'Nothing' and 'Stay' are by far the most frequent actions. Under this metric, a model that always predicts 'Nothing' for empty slots and 'Stay' for occupied slots could accumulate many correct entries without making any substantive scheduling decision. No majority-class or persistence baseline is reported. Please report per-slot and per-vessel accuracy, class-balanced metrics (e.g., macro-F1), and baselines such as always-Nothing, always-Stay, and previous-state persistence, before claiming that 82.64% reflects learned scheduling behavior.
  2. [Tables 4 and 5] All reported accuracies appear to come from a single run or split, with no error bars, confidence intervals, or repeated-seed experiments. Consequently, the small gaps between methods (e.g., Temporal-IRL 74.06% vs. XGBoost 75.89% for congestion, and 71.51% vs. 66.01% for leave accuracy) cannot be assessed for statistical significance. Please provide multiple seeds or resampling-based intervals and perform a significance test, or at minimum report the variance across runs.
  3. [Sections 3.1.1 and 3.2.1] Section 3.1.1 lists vessel origin, customs information, berth size, channel depth, and contractual agreements as factors that influence berth allocation, but Section 3.2.1 states that the implemented state features are only size, carrier, and stay time. The paper later claims that the learned reward function captures the port's underlying priorities and decision logic. This claim is not supported by the feature set. Please either temper the interpretation to describe the reward as conditional on the three modeled features, or add an ablation/sensitivity analysis that examines the effect of omitting the listed factors.
  4. [Section 3.3] The MaxEnt IRL description is generic, presenting Eqs. (2)-(6) without the implementation details needed for reproducibility. The paper does not specify the parameterization of the reward function R(s,a), how the partition function is computed or estimated over the 19-slot/10-action space, the transition model used in the Bellman equation, or the optimization hyperparameters (learning rate, number of iterations, discount factor). It is also unclear how the LSTM-AE features are combined with the one-hot state and action encodings before being passed to IRL. These details are essential; please provide a complete algorithmic description, pseudocode, or released code.
minor comments (7)
  1. [Abstract] The abstract contains several missing spaces, e.g., 'Improvedshipment planning, reducedelays' and 'optimizeinventoryanddistributionstrategies'; please proofread.
  2. [Table 2 and Section 3.1.2] Table 2's caption says the action space is 'a vector of length 19', but the text and Section 3.1.2 define the action vector as length 190. Please correct the inconsistency.
  3. [Figure 3] The caption lists 'Go to berth 4' twice and omits one of the berth actions in the enumeration; the action distribution itself also appears to show 'Go to berth 4' twice. Please fix the caption and verify the figure labels.
  4. [Eq. (1), Section 3.2.2] The loss is written as weighted binary cross-entropy over all action entries, but the action space is described as one-hot over 10 mutually exclusive actions per slot. Please clarify whether the model outputs a per-slot softmax and whether the loss is summed over slots, and specify the weights assigned to the non-downweighted actions.
  5. [Section 4.2.3] The congestion threshold (three or more vessels in the waiting area) is defined from observations in the same dataset (Figure 6). Please report sensitivity to this threshold or at least acknowledge the choice as dataset-derived, since the congestion accuracy metric depends directly on it.
  6. [Section 4.1] The temporal split is mentioned only as 'according to its temporal sequence'; please report the exact split dates, number of time windows in training and test sets, and the number of vessels/actions used in evaluation.
  7. [Section 4.2.2] The phrase 'state-of-the-art accuracy of 71.51%' refers to best among the five evaluated methods on one dataset; please qualify this as state-of-the-art among the compared baselines to avoid overclaiming.

Circularity Check

1 steps flagged · score 6.0 of 10

The 82.64% action accuracy is inflated by definitional 'Nothing' matches for empty slots; the central scheduling claim is not established by this metric.

  1. self definitional [Section 3.1.2 (States and Actions), Section 3.2.2, Section 4.2.2 Eq. (7)]
    "If a slot is empty, we encode it as a one-hot vector where all entries are 0, indicating "Nothing". ... accuracy = Number of matching actions / Total number of actions ... we found that the actions "Nothing," "Stay," and "Leave System" were significantly more frequent than other actions in the action space. Therefore, we substantially reduced the weight of the "Nothing" action and partially reduced the weights of the "Stay" and "Leave System" actions."

    An empty slot is, by definition, labeled 'Nothing,' and emptiness is directly visible in the state (IMO = 0, Section 3.1.2). Thus a model that maps empty slots to the all-zero 'Nothing' vector earns those matches without making any scheduling decision. Eq. (7) counts every such match in the reported 82.64% action-accuracy headline, while Section 3.2.2 concedes that 'Nothing' is overwhelmingly frequent and reduces its training weight to 0.1 but not its evaluation weight. The action-accuracy claim therefore reduces, in part, to the encoding definition rather than to a learned reward function; a trivial 'Nothing for empty slots, Stay for occupied slots' baseline would receive the same credit, and the paper reports no such baseline.

full rationale

The paper uses a temporal train/test split and the IRL reward is not fit on the test fold, so the general supervised-evaluation chain is not globally circular. The congestion and leave-accuracy metrics are also computed on held-out transitions and retain some independent content. However, the headline action-accuracy result (Table 4, Eq. 7) is partially circular by construction. Section 3.1.2 defines the action for any empty slot as the all-zero 'Nothing' vector, and emptiness is directly observable in the state via IMO = 0. A model that simply outputs 'Nothing' for empty slots and 'Stay' for occupied slots receives credit for all of those matches under Eq. (7). Section 3.2.2 concedes that 'Nothing' is the most frequent action and reduces its loss weight to 0.1 precisely for that reason, yet the evaluation counts it at full weight. No trivial majority-class or persistence baseline is reported, so the 82.64% figure cannot be read as evidence that the IRL reward learned berth-scheduling priorities. This is a real but partial circularity: the same definitional inflation does not directly determine the leave-accuracy number (71.51%), and the self-citations in the paper are background references rather than load-bearing. The stated limitations (single terminal, limited features, discretized windows) are acknowledged and do not themselves constitute circularity.

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

The model is built entirely from the Maher Terminal dataset; the state and action discretization, loss weights, and congestion threshold are all calibrated on the same data used for evaluation, which adds fitting risk. No external benchmarks or independent instrumentation are used, so the learned reward function has no outside validation. No new physical or conceptual entities are introduced beyond the standard IRL reward construct.

free parameters (5)
  • Time window (8 hours) = 8 hours
    Chosen as the midpoint of the dominant stay duration bin (Section 3.1.2, Figure 2); affects state and action definitions.
  • State slot counts = 6 berths + 7 waiting + 6 incoming = 19
    Set to the historical maximum number of vessels in each area observed in the same dataset (Section 3.1.2).
  • Loss weights for action classes = Nothing: 0.1; Stay and Leave System: 0.3
    Hand-set to counter class imbalance (Section 3.2.2); not tuned or justified.
  • Congestion threshold = 3 or more vessels waiting
    Derived from the observation in Figure 6 on the same dataset (Section 4.2.3); used in the congestion accuracy metric.
  • IRL and network hyperparameters = Not reported
    Discount factor, learning rate, LSTM hidden size, autoencoder dimension, and training epochs are not stated, despite being essential for reproduction.
assumptions (5)
  • domain assumption Expert scheduling behavior follows the maximum entropy softmax distribution over Q-values (Eq. 2).
    This is the standard MaxEnt IRL assumption, not proven from first principles for port operations.
  • domain assumption The port scheduling process is a Markov decision process over the discretized 19-slot state and 10-action space.
    Section 3.3 uses the Bellman equations, which require the Markov property and stationarity; real scheduling may depend on longer histories and exogenous events.
  • domain assumption AIS-derived trajectories correctly identify berth assignments, waiting states, and departure times.
    AIS data can be noisy or missing; the paper does not validate the reconstructed schedules against port records.
  • domain assumption The 19-slot state (6 berths, 7 waiting, 6 incoming) with features size, carrier, stay time is a sufficient statistic for the port's decision.
    The paper lists other decision factors (origin, customs, berth size, channel depth, contracts) in Section 3.1.1 but does not encode them in the state.
  • ad hoc to paper Removal of long empty periods does not change the scheduling policy being learned.
    Section 1.3 excludes weeks when the port was empty; if empty periods affect priority learning (e.g., recovery scheduling), the learned reward is biased. This exclusion is made for computational convenience without justification.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Temporal-IRL: Modeling Port Congestion and Berth Scheduling with Inverse Reinforcement Learning." pith.science (2026). https://pith.science/paper/WAINY5TW

@misc{pith2026250619843,
  author       = {Pith},
  title        = {Pith review of: Temporal-IRL: Modeling Port Congestion and Berth Scheduling with Inverse Reinforcement Learning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WAINY5TW}},
  note         = {Machine review of arXiv:2506.19843}
}
read the original abstract

Predicting port congestion is crucial for maintaining reliable global supply chains. Accurate forecasts enableimprovedshipment planning, reducedelaysand costs, and optimizeinventoryanddistributionstrategies, thereby ensuring timely deliveries and enhancing supply chain resilience. To achieve accurate predictions, analyzing vessel behavior and their stay times at specific port terminals is essential, focusing particularly on berth scheduling under various conditions. Crucially, the model must capture and learn the underlying priorities and patterns of berth scheduling. Berth scheduling and planning are influenced by a range of factors, including incoming vessel size, waiting times, and the status of vessels within the port terminal. By observing historical Automatic Identification System (AIS) positions of vessels, we reconstruct berth schedules, which are subsequently utilized to determine the reward function via Inverse Reinforcement Learning (IRL). For this purpose, we modeled a specific terminal at the Port of New York/New Jersey and developed Temporal-IRL. This Temporal-IRL model learns berth scheduling to predict vessel sequencing at the terminal and estimate vessel port stay, encompassing both waiting and berthing times, to forecast port congestion. Utilizing data from Maher Terminal spanning January 2015 to September 2023, we trained and tested the model, achieving demonstrably excellent results.

Review history (2 revisions) →

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

25 extracted references · 20 canonical work pages

  1. [1]

    A methodology to assess vessel berthing and speed optimization policies

    J Fernando Alvarez, Tore Longva, and Erna S Engebrethsen. A methodology to assess vessel berthing and speed optimization policies. Maritime economics & logistics, 12: 0 327--346, 2010

    work page 2010

  2. [2]

    Power relations in global supply chains and the unequal distribution of costs during crises: Abandoning garment suppliers and workers during the covid-19 pandemic

    Mark Anner. Power relations in global supply chains and the unequal distribution of costs during crises: Abandoning garment suppliers and workers during the covid-19 pandemic. International Labour Review, 161 0 (1): 0 59--82, 2022

    work page 2022

  3. [3]

    Elucidating us import supply chain dynamics

    Nikolay Aristov, Ziyan Li, Thomas Koch, and Elenna R Dugundji. Elucidating us import supply chain dynamics. Procedia Computer Science, 238: 0 216--223, 2024

    work page 2024

  4. [4]

    The cost of being landlocked: logistics costs and supply chain reliability, volume 4258

    Jean-Fran c ois Arvis, Gael Raballand, and Jean-Fran c ois Marteau. The cost of being landlocked: logistics costs and supply chain reliability, volume 4258. World Bank Publications, 2007

    work page 2007

  5. [5]

    Dynamic inverse reinforcement learning for characterizing animal behavior

    Zoe Ashwood, Aditi Jha, and Jonathan W Pillow. Dynamic inverse reinforcement learning for characterizing animal behavior. Advances in neural information processing systems, 35: 0 29663--29676, 2022

    work page 2022

  6. [6]

    Autoencoders

    Dor Bank, Noam Koenigstein, and Raja Giryes. Autoencoders. Machine learning for data science handbook: data mining and knowledge discovery handbook, pages 353--374, 2023

    work page 2023

  7. [7]

    The impact of port and trade security initiatives on maritime supply-chain management

    Ruth Banomyong*. The impact of port and trade security initiatives on maritime supply-chain management. Maritime Policy & Management, 32 0 (1): 0 3--13, 2005

    work page 2005

  8. [8]

    Using machine learning to predict port congestion: A study of the port of paranagu \'a

    Eirik Fadnes and Espen sheim Harviken. Using machine learning to predict port congestion: A study of the port of paranagu \'a . Master's thesis, 2023

    work page 2023

Show all 25 references
  1. [9]

    Predicting traffic phases from car sensor data using machine learning

    Emiliano Heyns, S Uniyal, Elenna Dugundji, F Tillema, and C Huijboom. Predicting traffic phases from car sensor data using machine learning. Procedia computer science, 151: 0 92--99, 2019

    2019

  2. [10]

    Investigation and imitation of human captains' maneuver using inverse reinforcement learning

    Takefumi Higaki, Hirotada Hashimoto, and Hitoshi Yoshioka. Investigation and imitation of human captains' maneuver using inverse reinforcement learning. 36: 0 137--148, 2022

    2022

  3. [11]

    Long short-term memory

    Sepp Hochreiter and J \"u rgen Schmidhuber. Long short-term memory. Neural computation, 9 0 (8): 0 1735--1780, 1997

    1997

  4. [12]

    Reinforcement learning: A survey

    Leslie Pack Kaelbling, Michael L Littman, and Andrew W Moore. Reinforcement learning: A survey. Journal of artificial intelligence research, 4: 0 237--285, 1996

    1996

  5. [13]

    Forecasting public transport ridership: Management of information systems using cnn and lstm architectures

    Sergey Khalil, Chintan Amrit, Thomas Koch, and Elenna Dugundji. Forecasting public transport ridership: Management of information systems using cnn and lstm architectures. Procedia Computer Science, 184: 0 283--290, 2021

    2021

  6. [14]

    Using artificial neural network model for berth congestion risk prediction

    NABIL LAMII, MOUHSENE FRI, CHARIF MABROUKI, et al. Using artificial neural network model for berth congestion risk prediction. IFAC-PapersOnLine, 55 0 (12): 0 592--597, 2022

    2022

  7. [15]

    Congestion analysis of waterborne, containerized imports from asia to the united states

    Robert C Leachman and Payman Jula. Congestion analysis of waterborne, containerized imports from asia to the united states. Transportation Research Part E: Logistics and Transportation Review, 47 0 (6): 0 992--1004, 2011

    2011

  8. [16]

    Algorithms for inverse reinforcement learning

    Andrew Y Ng, Stuart Russell, et al. Algorithms for inverse reinforcement learning. In Icml, volume 1, page 2, 2000

    2000

  9. [17]

    A deep learning approach for port congestion estimation and prediction

    Wenhao Peng, Xiwen Bai, Dong Yang, Kum Fai Yuen, and Junfeng Wu. A deep learning approach for port congestion estimation and prediction. Maritime Policy & Management, 50 0 (7): 0 835--860, 2023

    2023

  10. [18]

    Analysis of port waiting time due to congestion by applying markov chain analysis

    JFJ Pruyn, AA Kana, and WM Groeneveld. Analysis of port waiting time due to congestion by applying markov chain analysis. In Maritime Supply Chains, pages 69--94. Elsevier, 2020

    2020

  11. [19]

    Rumelhart, Geoffrey E

    David E. Rumelhart, Geoffrey E. Hinton, and Ronald J. Williams. Learning representations by back-propagating errors. Nature, 323 0 (6088): 0 533--536, 1986

    1986

  12. [20]

    Choosing a port: An analysis of containerized imports into the us

    Adams B Steven and Thomas M Corsi. Choosing a port: An analysis of containerized imports into the us. Transportation Research Part E: Logistics and Transportation Review, 48 0 (4): 0 881--895, 2012

    2012

  13. [21]

    The 2020 covid-19 pandemic and global value chains

    Roger Strange. The 2020 covid-19 pandemic and global value chains. Journal of Industrial and Business Economics, 47 0 (3): 0 455--465, 2020

    2020

  14. [22]

    Cross-lingual pitfalls: Automatic probing cross-lingual weakness of multilingual large language models, 2025 a

    Zixiang Xu, Yanbo Wang, Yue Huang, Xiuying Chen, Jieyu Zhao, Meng Jiang, and Xiangliang Zhang. Cross-lingual pitfalls: Automatic probing cross-lingual weakness of multilingual large language models, 2025 a . URL https://arxiv.org/abs/2505.18673

    2025 arXiv

  15. [23]

    Socialmaze: A benchmark for evaluating social reasoning in large language models, 2025 b

    Zixiang Xu, Yanbo Wang, Yue Huang, Jiayi Ye, Haomin Zhuang, Zirui Song, Lang Gao, Chenxi Wang, Zhaorun Chen, Yujun Zhou, Sixian Li, Wang Pan, Yue Zhao, Jieyu Zhao, Xiangliang Zhang, and Xiuying Chen. Socialmaze: A benchmark for evaluating social reasoning in large language mod...

    2025 arXiv

  16. [24]

    Evaluation of the marine traffic congestion of north harbor in busan port

    Gi-Tae Yeo, Michael Roe, and Sang-Moon Soak. Evaluation of the marine traffic congestion of north harbor in busan port. Journal of waterway, port, coastal, and ocean engineering, 133 0 (2): 0 87--93, 2007

    2007

  17. [25]

    Maximum entropy inverse reinforcement learning

    Brian D Ziebart, Andrew L Maas, J Andrew Bagnell, Anind K Dey, et al. Maximum entropy inverse reinforcement learning. In Aaai, volume 8, pages 1433--1438. Chicago, IL, USA, 2008

    2008

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.