Pith. sign in

REVIEW 3 major objections 4 minor 1 cited by

TransferLight: Zero-Shot Traffic Signal Control on any Road-Network

T0 review · 3 major / 4 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read TransferLight claims a single weight-tied policy, trained on randomized road networks, can control any unseen road network zero-shot, replacing pressure-based rewards with a symmetry-breaking log-distance reward.

desk verdict Worth a serious referee, but the 'any road-network' claim is overblown: the paper's own Ingolstadt21 run shows the zero-shot policy breaks on networks outside the training distribution. read the letter →

arxiv 2412.09719 v2 pith:RTOMJYNP submitted 2024-12-12 cs.AI cs.LG

Johann Schmidt , Frank Dreyer , Sayed Abid Hashimi , Sebastian Stober This is my paper · ORCID
classification cs.AIcs.LG
keywords trafficsignalcontrolzero-shottransferdomainrandomizationgraphneuralnetworkmulti-agentreinforcementlearninglog-distancerewardweight-tiedpolicydecentralized
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

TransferLight aims to show that traffic-signal control can be learned once and then applied, without retraining, to any road network: a single weight-tied policy, trained on randomly generated intersections with randomized traffic dynamics, is evaluated directly on unseen real-world benchmarks. The paper argues that the standard pressure-based reward is degenerate because arithmetic averaging makes it insensitive to where vehicles sit along a lane, and replaces it with a log-distance reward that breaks those symmetries. To handle arbitrary intersection geometries, it encodes the state with a hierarchical heterogeneous directed graph neural network that aggregates vehicles from lane segments to movements to phases, using a state-transition prior to give agents a view beyond their own intersection. If the claims hold, controllers would no longer need per-city training, and the same model could be dropped into new networks with different lane counts, lengths, phase sets, and traffic patterns. The paper itself notes the policy's limits on a 21-intersection scenario, where congestion builds after roughly 1200 steps.

What carries the argument

The central object is the log-distance reward, Eq. (5): $-\bigl|\sum_{(i,o)\in M_v}\frac{1}{|i|}\hat{E}_i - \frac{1}{|o|}\hat{E}_o\bigr|$, where $\hat{E}_\ell=\sum_{c\in C_\ell}\log(c+\epsilon)$ accumulates vehicle positions on a lane. Unlike pressure, which uses the arithmetic mean and is thus invariant to shuffling, shifting, or scaling vehicle positions, the log-domain energy breaks those symmetries, removing plateaus from the optimization surface. It is paired with a hierarchical, heterogeneous, directed graph encoder: lane segments (with densities, sinusoidal positional encodings, and a transition prior) aggregate into movement nodes, movements aggregate into phase nodes over a fully connected bipartite graph with edge features encoding prohibited/protected/permitted status, and an intra-level Jaccard-weighted propagation links phases; all agents share weights, so the same policy reads any intersection geometry and can be deployed on any network size.

What would settle it

Train TransferLight exactly as specified, then test it on a network whose lane lengths and departure-time distributions lie outside the sampled ranges, for example very long arterial blocks or double-peaked rush-hour profiles; if average travel time or queue length collapses relative to MaxPressure, the claim that the policy transfers to any road network is falsified, and the paper already gives a weaker version of this evidence on Ingolstadt21.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is a recipe for zero-shot generalisation in traffic signal control: train one weight-tied policy on domain-randomised road networks, and it transfers to unseen intersection geometries and traffic dynamics without re-training. The load-bearing mathematical piece is the log-distance reward, which the paper proves lifts the permutation, translation, and scale symmetries that make ordinary pressure-based rewards flat on large regions of the loss surface; replacing the density mean with cumulated log-distances makes the reward spatially aware. The architecture encodes each intersection as a directed heterogeneous graph through segment-to-movement and movement-to-phase attention, with a transition prior that lets agents anticipate flow from neighbouring intersections, and a global reward is shared across the decentralized agents. Empirically, TransferLight variants outperform or match trained and heuristic baselines on single-intersection benchmarks, Cologne8, Cologne3, and an arterial scenario, including baselines that were explicitly trained on those very scenarios. The authors also document a boundary: on Ingolstadt21, the zero-shot policy begins to congest after about 1200 steps and cannot recover, which they identify as a limitation to address in future work.

Load-bearing premise

The domain-randomization distribution over random road networks, lane lengths, and Beta-sampled departure times must be representative enough that a policy trained on it generalizes to any real-world network; if a target network's geometry or demand falls outside that training distribution, zero-shot performance degrades, as the paper's own Ingolstadt21 results show.

Editorial extensions

If this is right

  • A controller trained once on randomized networks can be deployed on a new city with no retraining or per-intersection tuning, provided the new network lies within the trained distribution.
  • The symmetry-breaking log-distance reward can replace pressure-based rewards in any intersection controller, giving spatially aware signal priorities even with heterogeneous lane lengths.
  • Because the state encoder reads lane segments rather than fixed-size vectors, new intersection layouts, phase sets, and lane counts can be handled without padding or architectural changes.
  • The weight-tied decentralized policy keeps inference cost per intersection constant as the network grows, since each agent is evaluated in parallel.
  • The documented failure on Ingolstadt21 sets a measurable boundary: zero-shot transfer works at the tested scales but not at 21 dense urban intersections under current training.

Reading between the lines

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

  • If the degeneracy analysis is correct, the same log-distance trick could be applied to other mean-based traffic metrics, such as queue length or waiting time, potentially removing plateaus without abandoning those well-studied objectives.
  • A testable extension is to quantify the generalization envelope by sweeping the domain-randomization ranges and recording zero-shot performance; the paper's Ingolstadt21 failure suggests a sharp phase transition in network size or demand intensity rather than a gradual decline.
  • The state-transition prior effectively gives each signal a one-hop lookahead; adding a similar prior over two or more hops, or over turn fractions, might extend the envelope to dense networks without retraining the encoder.
  • Because the policy is weight-tied and graph-structured, the same trained weights could in principle be fine-tuned on a new city with only a few steps of gradient updates, a hybrid between zero-shot transfer and full retraining that the paper does not explore.
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. TransferLight proposes a weight-tied, graph-based deep reinforcement learning controller for traffic signal control, trained exclusively on randomly generated road networks and traffic demands via domain randomization and then applied zero-shot to unseen benchmarks. Its main components are a log-distance reward intended as a symmetry-breaking alternative to pressure, a hierarchical heterogeneous directed GNN state encoder with segment-level positional encodings and transition priors, and DQN/A2C policy heads. Experiments compare TransferLight with heuristics and trained baselines on single-intersection benchmarks, Cologne8, Cologne3, and a synthetic arterial, while Appendix A.4 documents a failure on Ingolstadt21. The paper claims in the abstract and Section 1 that a single weight-tied policy scales zero-shot to any road network without re-training.

Significance. If the empirical results hold, the paper provides a useful demonstration of inductive, graph-based traffic signal control and a concrete architecture plus reward design that others can build on. The authors report open-sourced code, standard SUMO benchmarks, and a zero-shot protocol in which training is done only on random networks, which is a strength relative to baselines that are trained on the test network. However, the universal 'any road-network' claim is not supported by the evidence, and the theoretical degeneracy argument in Section 3 is informal. The empirical contribution on unseen benchmarks is the main strength; the universal generalization claim needs substantial qualification or additional evidence.

major comments (3)
  1. [Abstract, §1, §4, Appendix A.4] The central claim that a single weight-tied policy 'scales zero-shot to any road network' is contradicted by the paper's own Ingolstadt21 experiment: Fig. 9 shows both TL-DQN and TL-A2C diverging after roughly 1200 time steps, with congestion building and never recovering, and Section 7 concedes that the method needs further improvement for larger road networks. Because Ingolstadt21 is itself a road network, the universal quantifier in the abstract and the statement in Section 4 that the model 'can be applied to any road-network size' are empirically falsified by the authors' own benchmark. Please restrict the claim to the demonstrated regime, for example to networks sampled from the domain-randomization distribution and to comparable network scales, or provide systematic evidence that the Ingolstadt21 failure is an isolated artifact rather than a distributional boundary.
  2. [§3, Eqs. (2)–(5)] The 'proof' of degeneracies is not a proof and conflates two different gradient objects. The linearity properties of the mean show that the pressure reward is invariant under permutations of vehicle positions and equivariant under translations and scalings; they do not imply that the policy optimization loss has flat plateaus with zero gradients, because the policy gradient is taken with respect to the network parameters of an expected return, not with respect to the reward as a function of the state. This distinction matters because Section 3 motivates the log-distance reward as a fix for a claimed optimization failure. Please either provide a formal statement of the plateau claim or explicitly present this material as heuristic motivation.
  3. [§5, Eqs. (12)–(13)] The domain randomization distribution is described as containing 'all traffic-networks under some degree of freedom', but the actual sampling in the paper (uniform intersection counts and lane lengths, Beta-distributed departure times, and passenger cars only in training) has no demonstrated coverage of the kinds of real-world networks tested. In particular, Ingolstadt21 includes buses and a narrow urban topology, and Appendix A.4 shows exactly the kind of degradation one would expect from a train/test distribution shift. The paper does not quantify the distance between the training distribution and the test benchmarks, so the zero-shot generalization result is at best conditional on the training distribution. Please characterize the coverage of the training distribution and either expand it or weaken the transfer claim accordingly.
minor comments (4)
  1. [§3, Eq. (3)–(5)] The notation around the energy is inconsistent: after defining \hat{E}_\ell, the text defines E_\ell := C_\ell, and then Eqs. (4)–(5) appear to use \hat{E} without saying so explicitly; please align the notation.
  2. [§2, §6.2, figure captions] There are several typographical errors: 'Priliminaries' in the Section 2 header, 'conduced' in Section 6.2, and 'synthe\'c arterial scenario' in the captions of Figures 4, 5, and 9.
  3. [§6.1, Table 1 and Fig. 9] It would be helpful to state the number of random seeds for each result and to show confidence intervals or seed-wise spreads in the moving-average plots; the current presentation makes it difficult to assess the statistical reliability of the differences.
  4. [§4, Eqs. (10)–(11)] The sentence 'This changes Eq. (8) and Eq. (9) to' is imprecise because the modifications affect the attention scoring and node update formulas rather than Eqs. (8)–(9) themselves; please rephrase to refer to the movement-to-phase and intra-level propagation updates.

Circularity Check

0 steps flagged · score 2.0 of 10

No circular derivation: TransferLight's generalization claims rest on external benchmarks; the only self-citation is a non-load-bearing energy analogy.

full rationale

The paper's derivation chain is not circular. The log-distance reward (Eq. 5) is constructed from an explicit lane-energy definition (Eq. 3) and is compared against the standard pressure (Eq. 2); the lifted-degeneracy argument is presented in Section 3 with its own inequalities, and the reward is not fitted to the test benchmarks. The state-transition prior (Eq. 7) and the graph-neural-network encoder are design choices stated with explicit equations. The central zero-shot claim is evaluated on unseen external benchmarks (Cologne1/3/8, Ingolstadt1/7/21) against heuristic and independently trained baselines; no target-network parameter is fitted after training. Domain randomization (Section 5, Eqs. 12-13) samples a training distribution and does not incorporate test data. The only self-citation that touches the derivation is the energy analogy to (Schmidt, Köhler, and Borstell 2024) in Section 3; it is illustrative rather than load-bearing, since removing it leaves Eqs. (3)-(5) and the symmetry-breaking proof intact. A second self-citation (Schmidt and Stober 2024) appears only in future-work speculation. Finally, the paper's own Appendix A.4/Fig. 9 and Section 7 limitation show that the 'any road-network' claim is empirically too strong; that is an overgeneralization and correctness concern, not a circularity. Therefore no circular step is exhibited, and the score reflects only a minor non-load-bearing self-citation.

Assumptions & free parameters 3 free parameters · 5 assumptions · 1 invented entities

The central claims rest on a small number of hand-chosen hyperparameters and borrowed stability guarantees. The most consequential choices are the fixed segment length ds=10 m, the unspecified epsilon in the log-distance reward, and the domain randomization distribution. The paper provides no sensitivity analysis for ds or epsilon, and the log-distance energy is an ad hoc quantity without independent falsifiable evidence.

free parameters (3)
  • Segment length ds = 10 m
    Chosen by hand; controls spatial resolution of the segment-density encoding. Lanes shorter than ds are dropped, and no sensitivity analysis is provided. Its choice directly affects the state representation and the reward computation.
  • Epsilon in log-distance = ≈ 0 (not specified)
    The paper writes epsilon ≈ 0 without giving a numerical value. This parameter affects the magnitude of the log-distance reward and avoids log(0) at the intersection center.
  • Domain randomization distribution = alpha_f, beta_f ~ Unif(1,10); uniform sampling of intersections and lane lengths
    The distribution over training environments is a free choice that determines the generalization claim; if these distributions do not cover the test domain, zero-shot transfer fails, as seen on Ingolstadt21.
assumptions (5)
  • domain assumption Max-pressure control stabilizes traffic under mild assumptions (Wei et al. 2019a)
    The paper builds its reward design on these theoretical guarantees (Section 3) without re-deriving them.
  • domain assumption Segment-density representation with segment length ds fully describes the dynamical traffic system
    The paper asserts this in Section 4 and refers to Wei et al. (2019a) for the proof; it is a borrowed prior result.
  • domain assumption SUMO simulation is a faithful model of real traffic dynamics
    All experimental conclusions about generalization are drawn from SUMO simulations, so the sim-to-real transfer is assumed.
  • ad hoc to paper Log-distance energy E_l can be interpreted as a system energy whose minimization yields desirable traffic behavior
    This interpretation is introduced by the authors in Section 3 without independent justification; the actual reward is an absolute difference of log-energies, not the energy itself.
  • ad hoc to paper Equivalence classes of the pressure reward create zero-gradient plateaus that block gradient-based optimization
    Section 3 asserts this claim informally; it is not proved, and it ignores that the state encoder sees segment-level densities rather than the pressure value.
invented entities (1)
  • Log-distance energy E_l
    purpose: Reward shaping term to replace pressure-based density difference
    The paper introduces E_l = sum log(c+eps) as a new scalar associated with each lane. It has no falsifiable handle outside the paper; its value in improving control is only demonstrated on the paper's own benchmarks, without error bars.

how reviews work

0 comments
Cite this review

Pith. "Pith review of TransferLight: Zero-Shot Traffic Signal Control on any Road-Network." pith.science (2026). https://pith.science/paper/RTOMJYNP

@misc{pith2026241209719,
  author       = {Pith},
  title        = {Pith review of: TransferLight: Zero-Shot Traffic Signal Control on any Road-Network},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/RTOMJYNP}},
  note         = {Machine review of arXiv:2412.09719}
}
read the original abstract

Traffic signal control plays a crucial role in urban mobility. However, existing methods often struggle to generalize beyond their training environments to unseen scenarios with varying traffic dynamics. We present TransferLight, a novel framework designed for robust generalization across road-networks, diverse traffic conditions and intersection geometries. At its core, we propose a log-distance reward function, offering spatially-aware signal prioritization while remaining adaptable to varied lane configurations - overcoming the limitations of traditional pressure-based rewards. Our hierarchical, heterogeneous, and directed graph neural network architecture effectively captures granular traffic dynamics, enabling transferability to arbitrary intersection layouts. Using a decentralized multi-agent approach, global rewards, and novel state transition priors, we develop a single, weight-tied policy that scales zero-shot to any road network without re-training. Through domain randomization during training, we additionally enhance generalization capabilities. Experimental results validate TransferLight's superior performance in unseen scenarios, advancing practical, generalizable intelligent transportation systems to meet evolving urban traffic demands.

Figures

Figures reproduced from arXiv: 2412.09719 by the authors.

Figure 1
Figure 1. Our proposed traffic signal controller learns a gen [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Pressure (see Eq. (2)) is symmetric to vehicle po [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Our hierarchical state space encoding uses a [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: Test performances (moving averages) on Cologne8 over 3600 simulated time steps. Weight-Sharing Our universal state encoding function al￾lows using the model for each intersection. In this way, our model can be applied to any road-network size. Moreover, by sharing para…
Figure 5
Figure 5. Figure 5: Average Travel Time on Cologne3 over 3600 sim￾ulated time steps. popular traffic performance indicators during testing. We found that both TransferLight variants outperformed all heuristic and trained baselines. Note that, CoLight (Wei et al. 2019b) and SOTL (Reztsov 2…
Figure 6
Figure 6. Figure 6: Signal progression comparison on a synthetic 5- [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 7
Figure 7. Figure 7: Three random road-network samples (static envi [PITH_FULL_IMAGE:figures/full_fig_p010_7.png]
Figure 9
Figure 9. Figure 9: Test performances (moving averages) on Ingolstadt21 over 3600 simulated time steps. This is an example of the limits of generalisation capabilities of TransferLight. Around step 2400 a congestion is builds up around a few intersections which miscalculated some phase en…
Figure 10
Figure 10. Figure 10: Ablation study of positional encoding pe, edge features γmϕ and Jϕϕ′ , and our transition prior ρ¯ℓ comparing Average Travel Time (ATT) on cologne8. Left: visualization of the cologne8 network. Right: ablation results. Impact of Positional Encoding When lane segments …

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. A Graph-Based Control Interface for Traffic Signals on Heterogeneous Road Networks

    cs.LG 2026-07 accept novelty 6.0 of 10

    A shared movement-scoring graph network plus deterministic per-junction incidence matrices lets a single PPO policy execute on heterogeneous road networks with variable graph and action sizes.

Reference graph

Works this paper leans on

50 extracted references · 46 canonical work pages · cited by 1 Pith paper

  1. [1]

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

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

  2. [2]

    write newline

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

  3. [3]

    Brody, S.; Alon, U.; and Yahav, E. 2022. How Attentive are Graph Attention Networks? In International Conference on Learning Representations (ICLR)

    work page 2022

  4. [4]

    Chen, C.; Wei, H.; Xu, N.; Zheng, G.; Yang, M.; Xiong, Y.; Xu, K.; and Zhenhui. 2020. Toward A Thousand Lights: Decentralized Deep Reinforcement Learning for Large-Scale Traffic Signal Control. In AAAI Conference on Artificial Intelligence

    work page 2020

  5. [5]

    K.; Morales, P.; and Kochenderfer, M

    Choudhury, S.; Gupta, J. K.; Morales, P.; and Kochenderfer, M. J. 2021. Scalable Anytime Planning for Multi-Agent MDPs. In International Conference on Autonomous Agents and Multi-Agent Systems (AAMAS)

    work page 2021

  6. [6]

    Chu, T.; Wang, J.; Codec \`a , L.; and Li, Z. 2019. Multi-Agent Deep Reinforcement Learning for Large-Scale Traffic Signal Control. IEEE Transactions on Intelligent Transportation Systems

    work page 2019

  7. [7]

    Devailly, F.-X.; Larocque, D.; and Charlin, L. 2022. IG-RL: Inductive Graph Reinforcement Learning for Massive-Scale Traffic Signal Control. Trans. Intell. Transport. Sys

    work page 2022

  8. [8]

    Devailly, F.-X.; Larocque, D.; and Charlin, L. 2024. Model-Based Graph Reinforcement Learning for Inductive Traffic Signal Control. IEEE Open Journal of Intelligent Transportation Systems

    work page 2024

Show all 50 references
  1. [9]

    Dijkstra, E. W. 1959. A note on two problems in connexion with graphs. Numerische mathematik

    1959

  2. [10]

    v.; Guez, A.; and Silver, D

    Hasselt, H. v.; Guez, A.; and Silver, D. 2016. Deep reinforcement learning with double Q-Learning. In AAAI Conference on Artificial Intelligence

    2016

  3. [11]

    Jiang, H.; Li, Z.; Li, Z.; Bai, L.; Mao, H.; Ketter, W.; and Zhao, R. 2024. A General Scenario-Agnostic Reinforcement Learning for Traffic Signal Control. Trans. Intell. Transport. Sys

    2024

  4. [12]

    Z.; and Raileanu, R

    Jiang, Y.; Kolter, J. Z.; and Raileanu, R. 2024. On the importance of exploration for generalization in reinforcement learning. In International Conference on Neural Information Processing Systems (NeurIPS)

    2024

  5. [13]

    K.; Zhang, Y.; Bengio, Y.; and Ravanbakhsh, S

    Kaba, S.-O.; Mondal, A. K.; Zhang, Y.; Bengio, Y.; and Ravanbakhsh, S. 2023. Equivariance with Learned Canonicalization Functions. In International Conference on Machine Learning (ICML)

    2023

  6. [14]

    P.; and Ba, J

    Kingma, D. P.; and Ba, J. 2014. Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980

    2014 arXiv

  7. [15]

    N.; and Welling, M

    Kipf, T. N.; and Welling, M. 2017. Semi-Supervised Classification with Graph Convolutional Networks . In International Conference on Learning Representations (ICLR)

    2017

  8. [16]

    Korecki, M.; Dailisan, D.; and Helbing, D. 2023. How Well Do Reinforcement Learning Approaches Cope With Disruptions? The Case of Traffic Signal Control. IEEE Access

    2023

  9. [17]

    Little, J. D. C.; Kelson, M. D.; and Gartner, N. H. 1981. MAXBAND : a versatile program for setting signals on arteries and triangular networks . Working papers

    1981

  10. [18]

    Littman, M. L. 1994. Markov games as a framework for multi-agent reinforcement learning. In Cohen, W. W.; and Hirsh, H., eds., Machine Learning Proceedings

    1994

  11. [19]

    o tter \

    Lopez, P. A.; Behrisch, M.; Bieker-Walz, L.; Erdmann, J.; Fl \"o tter \"o d, Y.-P.; Hilbrich, R.; L \"u cken, L.; Rummel, J.; Wagner, P.; and Wie ner, E. 2018. Microscopic Traffic Simulation using SUMO. In IEEE International Conference on Intelligent Transportation Systems

    2018

  12. [20]

    Loshchilov, I.; and Hutter, F. 2017. Decoupled Weight Decay Regularization. In International Conference on Learning Representations (ICLR)

    2017

  13. [21]

    Mei, H.; Lei, X.; Da, L.; Shi, B.; and Wei, H. 2023. Libsignal: an open library for traffic signal control. Machine Learning

    2023

  14. [22]

    K.; Panigrahi, S

    Mondal, A. K.; Panigrahi, S. S.; Kaba, S.-O.; Rajeswar, S.; and Ravanbakhsh, S. 2023. Equivariant Adaptation of Large Pretrained Models. In Conference on Neural Information Processing Systems (NeurIPS)

    2023

  15. [23]

    Oroojlooy, A.; Nazari, M.; Hajinezhad, D.; and Silva, J. 2020. AttendLight: universal attention-based reinforcement learning model for traffic signal control. In International Conference on Neural Information Processing Systems (NeurIPS)

    2020

  16. [24]

    Peng, B.; Li, X.; Gao, J.; Liu, J.; Chen, Y.-N.; and Wong, K.-F. 2018. Adversarial Advantage Actor-Critic Model for Task-Completion Dialogue Policy Learning. In 2018 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

    2018

  17. [25]

    Reztsov, A. 2014. Self-Organising Traffic Lights (SOTL) as an Upper Bound Estimate. SSRN Electronic Journal, 24

    2014

  18. [26]

    Roess, R.; Prassas, E.; and McShane, W. 2004. Traffic engineering. Prentice Hall

    2004

  19. [27]

    Schmidt, J.; Köhler, B.; and Borstell, H. 2024. Reviving Simulated Annealing: Lifting its Degeneracies for Real-Time Job Scheduling. In Hawaii International Conference on System Sciences (HICSS)

    2024

  20. [28]

    Schmidt, J.; and Stober, S. 2024. Tilt your Head: Activating the Hidden Spatial-Invariance of Classifiers. In International Conference on Machine Learning (ICML)

    2024

  21. [29]

    Tobin, J.; Fong, R.; Ray, A.; Schneider, J.; Zaremba, W.; and Abbeel, P. 2017. Domain randomization for transferring deep neural networks from simulation to the real world. In IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)

    2017

  22. [30]

    Urbanik, T.; Tanaka, A.; Lozner, B.; Lindstrom, E.; Lee, K.; Quayle, S.; Beaird, S.; Tsoi, S.; Ryus, P.; Gettman, D.; Sunkari, S.; Balke, K.; and Bullock, D. 2015. NCHRP Report 812: A Guide for Applying Context-Sensitive Solutions for Signalized Intersections

    2015

  23. [31]

    A.; and Welling, M

    van der Pol, E.; van Hoof, H.; Oliehoek, F. A.; and Welling, M. 2022. Multi-Agent MDP Homomorphic Networks. In International Conference on Learning Representations (ICLR)

    2022

  24. [32]

    Varaiya, P. 2013. Max pressure control of a network of signalized intersections. Transportation Research Part C: Emerging Technologies

    2013

  25. [33]

    N.; Kaiser, L.; and Polosukhin, I

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

    2017

  26. [34]

    Veličković, P.; Cucurull, G.; Casanova, A.; Romero, A.; Liò, P.; and Bengio, Y. 2017. Graph Attention Networks. International Conference on Learning Representations (ICLR)

    2017

  27. [35]

    Wang, M.; Xiong, X.; Kan, Y.; Xu, C.; and Pun, M.-O. 2024. UniTSA: A Universal Reinforcement Learning Framework for V2X Traffic Signal Control. IEEE Transactions on Vehicular Technology

    2024

  28. [36]

    Wang, X.; Abdulhai, B.; and Sanner, S. 2023. A Critical Review of Traffic Signal Control and a Novel Unified View of Reinforcement Learning and Model Predictive Control Approaches for Adaptive Traffic Signal Control. In Handbook on Artificial Intelligence and Transport

    2023

  29. [37]

    Wang, Z.; Chen, J.; and Chen, H. 2021. EGAT: Edge-Featured Graph Attention Network. In Artificial Neural Networks and Machine Learning (ICANN)

    2021

  30. [38]

    Webster, F. 1958. Traffic Signal Settings. Road research technical paper

    1958

  31. [39]

    Wei, H.; Chen, C.; Zheng, G.; Wu, K.; Gayah, V.; Xu, K.; and Li, Z. 2019 a . PressLight: Learning Max Pressure Control to Coordinate Traffic Signals in Arterial Network. In ACM SIGKDD International Conference on Knowledge Discovery & Data Mining

    2019

  32. [40]

    Wei, H.; Xu, N.; Zhang, H.; Zheng, G.; Zang, X.; Chen, C.; Zhang, W.; Zhu, Y.; Xu, K.; and Li, Z. 2019 b . CoLight: Learning Network-level Cooperation for Traffic Signal Control. In ACM International Conference on Information and Knowledge Management (CIKM)

    2019

  33. [41]

    Wei, H.; Zheng, G.; Gayah, V.; and Li, Z. 2021. Recent Advances in Reinforcement Learning for Traffic Signal Control: A Survey of Models and Evaluation. SIGKDD Explor. Newsl

    2021

  34. [42]

    V.; and Li, Z

    Wei, H.; Zheng, G.; Gayah, V. V.; and Li, Z. J. 2019 c . A Survey on Traffic Signal Control Methods. ArXiv, abs/1904.08117

    2019 arXiv

  35. [43]

    Wei, H.; Zheng, G.; Yao, H.; and Li, Z. 2018. IntelliLight: A Reinforcement Learning Approach for Intelligent Traffic Light Control. In ACM SIGKDD International Conference on Knowledge Discovery & Data Mining

    2018

  36. [44]

    Wu, C.; Kim, I.; and Ma, Z. 2023. Deep Reinforcement Learning Based Traffic Signal Control: A Comparative Analysis. Procedia Computer Science. International Conference on Ambient Systems, Networks and Technologies Networks (ANT) and International Conference on Emerging Data an...

    2023

  37. [45]

    Yi, Y.; Li, G.; Wang, Y.; and Lu, Z. 2024. Learning to share in multi-agent reinforcement learning. In International Conference on Neural Information Processing Systems (NeurIPS)

    2024

  38. [46]

    Yoon, J.; Ahn, K.; Park, J.; and Yeo, H. 2021. Transferable traffic signal control: Reinforcement learning with graph centric state representation. Transportation Research Part C: Emerging Technologies

    2021

  39. [47]

    Yu, R.; Wang, Z.; Wang, Y.; Li, K.; Liu, C.; Duan, H.; Ji, X.; and Chen, J. 2023. LaPE: Layer-adaptive Position Embedding for Vision Transformers with Independent Layer Normalization. In IEEE/CVF International Conference on Computer Vision (ICCV)

    2023

  40. [48]

    Zang, X.; Yao, H.; Zheng, G.; Xu, N.; Xu, K.; and Li, Z. 2020. MetaLight: Value-based Meta-reinforcement Learning for Traffic Signal Control. In AAAI Conference on Artificial Intelligence

    2020

  41. [49]

    Zhang, Z.; Yang, J.; and Zha, H. 2020. Integrating Independent and Centralized Multi-agent Reinforcement Learning for Traffic Signal Network Optimization. In International Conference on Autonomous Agents and Multi-Agent Systems (AAMAS)

    2020

  42. [50]

    Zheng, G.; Xiong, Y.; Zang, X.; Feng, J.; Wei, H.; Zhang, H.; Li, Y.; Xu, K.; and Li, Z. 2019. Learning Phase Competition for Traffic Signal Control. In ACM International Conference on Information and Knowledge Management (CIKM)

    2019

Pith tools

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