DeepIST: Deep Image-based Spatio-Temporal Network for Travel Time Estimation

Tao-yang Fu; Wang-Chien Lee

REVIEW 5 major objections 4 minor 38 references

DeepIST: Deep Image-based Spatio-Temporal Network for Travel Time Estimation

T0 review · 5 major / 4 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read DeepIST claims that representing a route as a sequence of map-like images and processing them with CNNs reduces travel-time mean absolute error by 24.37% on Porto and 25.64% on Chengdu.

desk verdict A genuinely new image-sequence architecture for travel time estimation, but the headline 24-25% MAE gain is vulnerable to a missing control on the traffic-condition channel. read the letter →

arxiv 1909.05637 v1 pith:EVJKRJAU submitted 2019-09-05 cs.CV cs.LGstat.ML

Tao-yang Fu , Wang-Chien Lee This is my paper · ORCID

classification cs.CVcs.LGstat.ML

keywords traveltimeestimationconvolutionalneuralnetworksspatio-temporaldatatrajectorymininggeneralizedimagespathrepresentationroaddeeplearning

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 DeepIST, a neural network that estimates the travel time of a given route by turning the route into a sequence of map-like images and processing those images with convolutional networks. The authors argue that prior methods fail because they use hand-crafted features or shallow networks that cannot capture the spatial patterns of movement along a path, and because recurrent temporal models suffer gradient problems. DeepIST instead splits a path into overlapping sub-path windows, plots each window as a multi-channel image containing the sub-path, estimated traffic condition, road network, and traffic signals, then uses a specially regularized 2D CNN (PathCNN) to extract spatial patterns and a 1D CNN to combine them over time. On two large taxi datasets, Porto and Chengdu, the paper reports mean absolute error reductions of 24.37% and 25.64% over the best existing models, and ablations attribute the gain to the image representation, the line-focused regularization, and the 1D temporal layer.

What carries the argument

The generalized image is the central object: a fixed-resolution $k\times k$ tensor whose channels hold the sub-path, estimated traffic-condition values, the road network, and traffic signals, so that ordinary CNN machinery can be applied to route geometry. PathCNN is the mechanism that makes the representation work for lines: it runs two convolutions per block in parallel, one followed by max pooling for binary structure and one by average pooling for numeric traffic values, and adds a penalty that rewards a large center element and diverse off-center elements in each filter, steering filters toward centered line detectors. The temporal layer is a 1D CNN stacked over the sequence of spatial pattern vectors, which captures local ordering without recurrent gradient problems.

What would settle it

Retrain or reconstruct the traffic-condition channel using only the training split of each dataset, rerun DeepIST against the same baselines, and compare MAE; if the 24.37% and 25.64% margins disappear or shrink sharply, the reported gain depended on data leakage rather than on the image-based architecture.

Watch

Extended reading notes

Core claim

The central claim is that path-level travel time can be accurately estimated by representing a path as a sequence of generalized images rather than as feature vectors or point sequences. In DeepIST, a sliding window cuts the path into overlapping half-kilometer sub-paths; each sub-path is plotted into a 100x100 pixel image with separate channels for the sub-path itself, the hourly traffic condition of the road segments, the underlying road network, and nearby traffic signals. A new 2D CNN, PathCNN, extracts spatial moving patterns from each image using parallel max and average pooling branches, plus three penalties that push convolution filters to detect lines at the center of their receptive field. The resulting sequence of spatial pattern vectors is passed through a 1D CNN that captures local temporal dependencies, and the network is trained end-to-end with a multi-task loss that also predicts sub-path travel times. The paper reports that this architecture outperforms the best previous methods by 24.37% in MAE on Porto and 25.64% on Chengdu, and its ablation study shows the traffic-condition and road-network channels, overlapping windows, and the line regularization each contribute to the gain.

Load-bearing premise

The load-bearing premise is that the hourly traffic-condition channel is produced by an external speed model trained only on the training split, and the paper does not demonstrate that, so a violation would let DeepIST see test-period information the baselines do not get.

Editorial extensions

If this is right

If the reported results hold, travel time estimation can move from manual feature engineering to an end-to-end image-based model that improves accuracy by roughly a quarter on large real-world taxi datasets.
Because the image representation has an open-ended channel dimension, adding further factors such as weather, speed limits, or driver behavior is a direct extension within the same architecture.
The line-focused regularization in PathCNN can be reused in any CNN task where the objects of interest are thin structures such as roads, vessels, or text rather than textures.
The finding that a 1D CNN outperforms an LSTM on the same sequence of spatial patterns suggests that local temporal dependencies, rather than long-range recurrence, carry most of the signal for travel time estimation.

Reading between the lines

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

A strict test would rebuild the traffic-condition channel using only the training split, then rerun the comparison; if the 24-25% margin shrinks, part of the reported gain is leakage rather than architecture.
The same path-to-image-sequence treatment should transfer to other route-level prediction tasks, such as estimated fuel use or delivery delay, whenever route geometry plus environmental context matters.
The relative contribution of each channel could be probed by attention or saliency maps on the learned images; the paper's own ablations suggest traffic condition and road network carry most of the signal, while traffic signals add little, possibly because public map data on signals is sparse.
If the 1D-CNN temporal layer generalizes, it suggests replacing recurrent layers with convolutional sequence models in other spatio-temporal regression tasks, avoiding vanishing-gradient training difficulties.

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.

Desk Editor's Note

read the letter

Colleague,

The thing to know: this is a real architecture paper, not an incremental hyperparameter sweep. The core move — slicing a path into overlapping sub-paths, plotting each into a multi-channel "generalized image" (sub-path, traffic condition, road network, signals), then running a customized 2D CNN (PathCNN) followed by a 1D CNN over the sequence — is genuinely new for travel time estimation. PathCNN's line-oriented regularization (center activation, diversity penalty, L2) is a sensible response to the known texture-bias of CNNs, and the dual max/avg pooling is justified by the heterogeneous channels. The idea is clean and the experiments are substantial: two large real taxi datasets, five baselines spanning road/path/learning methods, and sensitivity analysis on window size, image resolution, layer counts, and training set size. The reported 24-25% MAE improvement is at least plausible as an architectural effect.

The soft spot that matters is the traffic-condition channel. In Section 4.1 the authors say they use a state-of-the-art spd-LSTM method [20] to estimate hourly traffic conditions per road segment, normalized by the maximum speed of the whole dataset. They never state whether that spd-LSTM was trained only on the training split. Because the data split is random (80/10/10), test trips share road segments and hours with training trips. If the hourly traffic-condition estimates were fit on the full dataset, the traffic channel can encode average speeds — essentially the target — for the very trips being predicted. Figure 10(a) shows that adding this channel gives a large boost over sub-path-only. That makes the headline improvement vulnerable to leakage. This is directly testable: rerun with a traffic-condition model trained only on the training split, or ablate the channel entirely.

Smaller soft spots: there are no error bars or significance tests across the five runs; Table 2 reports only means. The sub-path ground truths come from a constant-speed assumption on map-matched trajectories — that is a modeling choice, not circularity, as the stress-test reader correctly notes. Baselines like WDR and DeepTTE may not have access to the same traffic-condition input, so the comparison is not perfectly controlled. No code or data is released, which makes replication harder. The citation pattern is ordinary and appropriate.

Who is this for? Anyone working on deep learning for trajectory speed and travel-time estimation. It deserves a serious referee — the architecture idea is worth engaging with even if the empirical claim needs tightening. If I were the editor, I would send it out with a specific request to address the leakage control and error bars.

Candidly, I'd want the leakage issue resolved before trusting the quantitative headline, but the paper is a genuine contribution.

Referee Report

5 major / 4 minor

Summary. The paper proposes DeepIST, a neural network for travel time estimation of a given path. A path is divided into overlapping sub-paths via a sliding window; each sub-path is rasterized into a multi-channel 'generalized image' containing the sub-path, estimated traffic condition, road network, and traffic signals. A 2D CNN (PathCNN) with two parallel pooling branches and line-oriented regularization extracts spatial features, and a 1D CNN captures temporal dependencies among the resulting feature sequence. The model is trained with a multi-task loss that also predicts sub-path travel times, where sub-path ground truth is derived from a constant-speed assumption between consecutive GPS sample points. On Porto and Chengdu taxi datasets, the authors report 24.37% and 25.64% MAE improvements over the best baselines.

Significance. If the comparison were fully controlled, the generalized-image representation and the 1D-CNN temporal layer would be a plausible and useful contribution: it offers a way to apply CNN inductive biases to path-based prediction, and the ablation studies (Fig. 10) individually examine the contributions of channels, window overlap, pooling design, multi-task learning, and regularization. The empirical gains are large and the paper gives a detailed parameter sensitivity analysis. However, the central quantitative claim is currently not established because of a potential leakage path through the externally generated traffic-condition channel and because the comparison gives DeepIST input information not available to baselines. The paper does not provide code, variance estimates, or significance tests, so the claimed margins cannot be independently assessed. With those controls in place the contribution would be significant for the spatio-temporal data mining community.

major comments (5)

[4.1, 5.3] The traffic-condition channel is generated by an external spd-LSTM model [20] and normalized by the maximum speed of the whole dataset (Section 4.1). The paper does not state whether the external model was trained only on the training split. Because the 80/10/10 split is random (Section 5.3), test trajectories share road segments, hours, and even individual trips with the training data; if the traffic model or the normalization statistic used any test-period information, the traffic channel can encode the average speed of the very trips being predicted. Figure 10(a) shows that adding this channel (P+T) is responsible for a large improvement, so the reported 24.37% and 25.64% MAE gains in Table 2 may be an artifact of leakage rather than of PathCNN or the 1D-CNN. Please specify the training protocol for the traffic model, ensure it uses only the training split (including for normalization), or re-run the comparison with traffic features made available to all baselines.
[5.2, Table 2] Even without leakage, the comparison in Table 2 is not controlled: DeepIST receives an additional input channel (hourly traffic condition) that the learning-based baselines DeepTravel, WDR, and DeepTTE do not appear to receive. Since Figure 10(a) demonstrates that this channel alone yields a large part of the improvement, the reported advantage over 'the best existing models' conflates additional input information with the proposed architecture. Please include a DeepIST variant without the traffic channel (P-only in Figure 10(a)) alongside baselines augmented with equivalent traffic-condition features, or otherwise factor out the value of the extra channel.
[5.3, Table 2, Figs. 8-10] Section 5.3 states that each experiment is repeated 5 times and the mean is reported, but no standard deviation, confidence interval, or significance test is given in Table 2 or in the sensitivity and ablation figures. Without a measure of variance, the 24-25% MAE improvements and the ordering of the methods cannot be statistically distinguished from noise, especially for configurations that are close (e.g., the claimed advantage of DeepISTLST M over WDR on Porto). Please report per-run results or error bars and, where relevant, paired significance tests.
[4.2, Eq. (1)] The diversity penalty appears to have the opposite sign to its stated purpose. The text says the values of the non-center elements should be 'diverse (i.e., not all of them are similar)', but L_div is defined as -Σ H(δ(c \ c.center)), where H is Shannon entropy. Since Shannon entropy is maximized when the probability distribution is uniform, minimizing -H encourages the softmax-normalized non-center values to be equal, i.e., mutually similar. Please correct the sign/definition or clarify the intended behavior; as written, this penalty does not implement the line-oriented diversity criterion described in the text.
[5.4, Table 2] The observation that 'DeepISTLST M achieves the best performance' among DeepTravel, WDR, DeepTTE and DeepISTLST M is contradicted by Table 2 on Porto: WDR has MAE 70.67 while DeepISTLST M has MAE 95.29 (though DeepISTLST M has lower MAPE). On Chengdu, DeepISTLST M is better on all metrics. Please reconcile this statement with the reported numbers or clarify which metric supports the claim.

minor comments (4)

[5.3, 5.6] Section 5.3 sets the default sliding step s=0.4 km, while Section 5.6 describes the 'Best' setting as w=0.5 km and s=0.1 km; please clarify which value was used for Table 2 and Fig. 9.
[5.5, 5.6] There are typos: 'desribing' and 'respectivly' in Section 5.5, and 'One the other hand' in Section 5.6.
[Figure 5] The figure caption shows the input tensor as 100×100×3 while the text says the images have d=4 channels; please clarify whether the figure is illustrative or update it to match the actual input.
[5.2, 5.3] The baseline hyperparameters are described only as 'tuned to the best parameter settings'; for reproducibility, please report the search ranges and final selected hyperparameters for DeepTravel, WDR, DeepTTE, and DeepISTLST M.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: DeepIST's travel-time output is a standard supervised regression on ground-truth labels; none of the auxiliary constructions reduces to the target by definition or by self-citation.

full rationale

The paper's derivation chain is self-contained and non-circular. DeepIST predicts travel time from path-derived generalized images, and the model is trained end-to-end by minimizing MAPE against ground-truth path travel times (Eq. 3). The sub-path supervision in Eq. 4 is constructed from the same trajectories via a constant-speed interpolation assumption; this is an auxiliary training target, not a fitted parameter later reported as a prediction. The PathCNN penalties in Eq. 1 encode modeling preferences about line-centered, diverse convolution filters; they are regularizers, not hidden restatements of the output. The traffic-condition channel is produced by an external method [20] and then normalized; while the paper does not specify the training split used to generate those hourly estimates, any leakage there would be an experimental-control or data-leakage problem, not a circular derivation. There are no load-bearing self-citations: the references to prior work by others are external, and the comparison baselines are independently implemented and tuned. The claimed improvements over baselines are empirical benchmark results, not consequences of the paper's definitions. Therefore no step in the claimed derivation reduces to its own inputs, and the circularity score is 0.

Assumptions & free parameters 9 free parameters · 5 assumptions · 0 invented entities

The central claim depends on standard supervised learning with a large number of hand-tuned hyperparameters and a potentially leaky traffic-condition feature. No new physical entities are introduced.

free parameters (9)

window size w = 0.5 km
Sub-path length plotted into each image; tuned via sensitivity analysis (Fig 9e).
sliding step s = 0.4 km
Overlap between adjacent windows; chosen by hand, not fully tuned.
image size k x k = 100 x 100
Resolution of each generalized image; tuned via sensitivity analysis (Fig 9f).
number of PathCNN layers M = 4
Depth of spatial CNN; tuned via sensitivity analysis (Fig 9a).
number of 1D-CNN layers N = 2
Depth of temporal CNN; tuned via sensitivity analysis (Fig 9c).
number of convolutions in PathCNN = 16
Width of spatial CNN; selected based on Fig 9b.
number of convolutions in 1D-CNN = 1024
Width of temporal CNN; selected based on Fig 9d.
loss weights beta, gamma1, gamma2, gamma3 = 0.6, 0.1, 0.1, 0.01
Weights for path loss, sub-path loss, center penalty, diversity penalty, and L2 penalty; chosen empirically (Section 5.3, 5.5).
Smax = 53 for Porto
Max sequence length of spatial patterns; set to cover more than 99% of sequences in the data.

assumptions (5)

domain assumption Traffic condition can be estimated from historical data using the spd-LSTM method [20] and is independent of the target trajectory's travel time.
If the traffic-condition model is trained on data that includes the test trajectories, this input leaks target information; the paper does not specify the training split for this external model (Section 4.1).
domain assumption Map-matching results are accurate enough for image generation and sub-path labeling.
The entire pipeline depends on trajectories being correctly mapped to road segments via [21,22] (Section 5.1).
domain assumption Moving speed between consecutive GPS sample points is constant, allowing sub-path travel times to be derived.
Used to create ground truth for the multi-task sub-path loss (Section 4.4).
domain assumption Tobler's First Law of Geography ('near things are more related than distant things') justifies using 1D-CNN rather than LSTM for temporal patterns.
Cited as the intuition for the temporal layer design (Section 4.3).
ad hoc to paper Rasterizing paths and contextual information into fixed-size images preserves the information needed for travel time estimation.
The whole input representation depends on this; window size, image resolution, and channel choices are not derived from theory.

how reviews work

0 comments

Cite this review

Pith. "Pith review of DeepIST: Deep Image-based Spatio-Temporal Network for Travel Time Estimation." pith.science (2026). https://pith.science/paper/EVJKRJAU

@misc{pith2026190905637,
  author       = {Pith},
  title        = {Pith review of: DeepIST: Deep Image-based Spatio-Temporal Network for Travel Time Estimation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/EVJKRJAU}},
  note         = {Machine review of arXiv:1909.05637}
}

read the original abstract

Estimating the travel time for a given path is a fundamental problem in many urban transportation systems. However, prior works fail to well capture moving behaviors embedded in paths and thus do not estimate the travel time accurately. To fill in this gap, in this work, we propose a novel neural network framework, namely {\em Deep Image-based Spatio-Temporal network (DeepIST)}, for travel time estimation of a given path. The novelty of DeepIST lies in the following aspects: 1) we propose to plot a path as a sequence of "generalized images" which include sub-paths along with additional information, such as traffic conditions, road network and traffic signals, in order to harness the power of convolutional neural network model (CNN) on image processing; 2) we design a novel two-dimensional CNN, namely {\em PathCNN}, to extract spatial patterns for lines in images by regularization and adopting multiple pooling methods; and 3) we apply a one-dimensional CNN to capture temporal patterns among the spatial patterns along the paths for the estimation. Empirical results show that DeepIST soundly outperforms the state-of-the-art travel time estimation models by 24.37\% to 25.64\% of mean absolute error (MAE) in multiple large-scale real-world datasets.

Figures

Figures reproduced from arXiv: 1909.05637 by the authors.

**Figure 2.** The architecture of DeepIST really complex. Therefore, research on developing effective techniques to exploit the abundant collected trajectory data available nowadays is a mandate. Ideally, an effective travel time estimation technique would automatically learn important moving patterns (in both of the spatial and temporal dimensions) along paths corresponding to various major factors, leading to accurate estimat… view at source ↗

**Figure 3.** Data preparation layer [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗

Figures from the paper (4 more)

**Figure 6.** Figure 6: Regularizations for Paths size of the image and d is the number of channels used in the image (e.g., k=100 and d=4). PathCNN takes Ii as input x 0 i and feeds x 0 i into M max+avg layers. In each max+avg layer m, PathCNN has two separate convolutional layers, which are…

**Figure 7.** Figure 7: Temporal Layer sliding window to capture the local temporal patterns among spatial patterns in each window, and use the later convolutional layers further capture higher-level temporal patterns hierarchically. Model Architecture. As shown in [PITH_FULL_IMAGE:figures/…

**Figure 8.** Figure 8: Training Set Size patterns while exploring different approaches to extract spatial moving patterns, DeepISTLST M achieves the best performance, indicating that it is more effective than others in extracting spatial moving patterns by exploring the novel ideas in PathC…

**Figure 10.** Figure 10: Comparison of approaches to issues in DeepIST [PITH_FULL_IMAGE:figures/full_fig_p009_10.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

38 extracted references · 30 canonical work pages

[20]

Xiaolei Ma, Zhimin Tao, Yinhai Wang, Haiyang Yu, and Yunpeng Wang. 2015. Long short-term memory neural network for traffic speed prediction using remote microwave sensor data. Transportation Research Part C: Emerging Technologies 54 (2015), 187–197

2015
[1]

OpenStreetMap

2004. OpenStreetMap. https://www.openstreetmap.org/

work page 2004
[2]

Taxi Service Trajectory (TST) Prediction Challenge

2015. Taxi Service Trajectory (TST) Prediction Challenge. http://www.geolink. pt/ecmlpkdd2015-challenge/

work page 2015
[3]

Taxi Travel Time Prediction Challenge

2016. Taxi Travel Time Prediction Challenge. http://www.dcjingsai.com

work page 2016
[4]

Muhammad Tayyab Asif, Justin Dauwels, Chong Yang Goh, Ali Oran, Esmail Fathi, Muye Xu, Menoth Mohan Dhanya, Nikola Mitrovic, and Patrick Jaillet. 2014. Spatiotemporal patterns in large-scale traffic speed prediction. IEEE Transactions on Intelligent Transportation Systems 15, 2 (2014), 794–804

work page 2014
[5]

Corrado De Fabritiis, Roberto Ragona, and Gaetano Valenti. 2008. Traffic esti- mation and prediction based on real time floating car data. In International IEEE Conference on Intelligent Transportation Systems . IEEE, 197–203

work page 2008
[6]

Robert Geirhos, Patricia Rubisch, Claudio Michaelis, Matthias Bethge, Felix A Wichmann, and Wieland Brendel. 2018. ImageNet-trained CNNs are biased towards texture; increasing shape bias improves accuracy and robustness. arXiv preprint arXiv:1811.12231 (2018)

arXiv 2018
[7]

Klaus Greff, Rupesh K Srivastava, Jan Koutník, Bas R Steunebrink, and Jürgen Schmidhuber. 2017. LSTM: A search space odyssey. IEEE transactions on neural networks and learning systems 28, 10 (2017), 2222–2232

2017

Show all 38 references

[8]

Geoffrey E Hinton, Nitish Srivastava, Alex Krizhevsky, Ilya Sutskever, and Ruslan R Salakhutdinov. 2012. Improving neural networks by preventing co- adaptation of feature detectors. arXiv preprint arXiv:1207.0580 (2012)

2012 arXiv
[9]

Aude Hofleitner, Ryan Herring, Pieter Abbeel, and Alexandre Bayen. 2012. Learn- ing the dynamics of arterial traffic from probe data using a dynamic Bayesian network. IEEE Transactions on Intelligent Transportation Systems 13, 4 (2012)

2012
[10]

Saihui Hou and Zilei Wang. 2019. Weighted Channel Dropout for Regularization of Deep Convolutional Neural Network. (2019)

2019
[11]

Sergey Ioffe and Christian Szegedy. 2015. Batch normalization: Accelerating deep network training by reducing internal covariate shift. arXiv preprint arXiv:1502.03167 (2015)

2015 arXiv
[12]

Erik Jenelius and Haris N Koutsopoulos. 2013. Travel time estimation for urban road networks using low frequency probe vehicle data. Transportation Research Part B: Methodological 53 (2013), 64–81

2013
[13]

Zhanfeng Jia, Chao Chen, Ben Coifman, and Pravin Varaiya. 2001. The PeMS algorithms for accurate, real-time estimates of g-factors and speeds from single- loop detectors. In Intelligent Transportation Systems. IEEE, 536–541

2001
[14]

Yoon Kim. 2014. Convolutional neural networks for sentence classification.arXiv preprint arXiv:1408.5882 (2014)

2014 arXiv
[15]

Diederik P Kingma and Jimmy Ba. 2014. Adam: A method for stochastic opti- mization. arXiv preprint arXiv:1412.6980 (2014)

2014 arXiv
[16]

Alex Krizhevsky, Ilya Sutskever, and Geoffrey E Hinton. 2012. Imagenet classifica- tion with deep convolutional neural networks. In Advances in neural information processing systems. 1097–1105

2012
[17]

Yaguang Li, Kun Fu, Zheng Wang, Cyrus Shahabi, Jieping Ye, and Yan Liu. 2018. Multi-task representation learning for travel time estimation. InProceedings of the ACM SIGKDD international conference on Knowledge discovery and data mining . ACM, 1695–1704

2018
[18]

Wuman Luo, Haoyu Tan, Lei Chen, and Lionel M Ni. 2013. Finding time period- based most frequent path in big trajectory data. In Proceedings of the ACM SIG- MOD international conference on management of data . ACM, 713–724

2013
[19]

Yisheng Lv, Yanjie Duan, Wenwen Kang, Zhengxi Li, Fei-Yue Wang, et al. 2015. Traffic flow prediction with big data: A deep learning approach.IEEE Transactions Intelligent Transportation Systems 16, 2 (2015), 865–873

2015
[21]

Sebastian Mattheis. 2016. Barefoot. https://github.com/bmwcarit/barefoot/

2016
[22]

Paul Newson and John Krumm. 2009. Hidden Markov map matching through noise and sparseness. In Proceedings of the ACM SIGSPATIAL international confer- ence on advances in geographic information systems . ACM, 336–343

2009
[23]

Hamid Palangi, Li Deng, Yelong Shen, Jianfeng Gao, Xiaodong He, Jianshu Chen, Xinying Song, and Rabab Ward. 2016. Deep sentence embedding using long short-term memory networks: Analysis and application to information retrieval. IEEE/ACM Transactions on Audio, Speech and Langua...

2016
[24]

Razvan Pascanu, Tomas Mikolov, and Yoshua Bengio. 2013. On the difficulty of training recurrent neural networks. In International Conference on Machine Learning. 1310–1318

2013
[25]

Mahmood Rahmani, Erik Jenelius, and Haris N Koutsopoulos. 2013. Route travel time estimation using low-frequency floating car data. In Proceedings of the International IEEE Conference on Intelligent Transportation Systems . IEEE

2013
[26]

John Rice and Erik Van Zwet. 2004. A simple and effective method for predicting travel times on freeways. IEEE Transactions on Intelligent Transportation Systems 5, 3 (2004), 200–207

2004
[27]

Nitish Srivastava, Elman Mansimov, and Ruslan Salakhudinov. 2015. Unsu- pervised learning of video representations using lstms. In Proceedings of the International conference on machine learning . 843–852

2015
[28]

Ilya Sutskever, Oriol Vinyals, and Quoc V Le. 2014. Sequence to sequence learning with neural networks. In Advances in neural information processing systems

2014
[29]

Waldo R Tobler. 1970. A computer movie simulating urban growth in the Detroit region. Economic geography 46, sup1 (1970), 234–240

1970
[30]

Dong Wang, Wei Cao, Mengwen Xu, and Jian Li. 2016. Etcps: An effective and scalable traffic condition prediction system. In International Conference on Database Systems for Advanced Applications . Springer, 419–436

2016
[31]

Dong Wang, Junbo Zhang, Wei Cao, Jian Li, and Yu Zheng. 2018. When Will You Arrive? Estimating Travel Time Based on Deep Neural Networks. Proceedings of the AAAI National Conference on Artificial Intelligence

2018
[32]

Hongjian Wang, Yu-Hsuan Kuo, Daniel Kifer, and Zhenhui Li. 2016. A simple baseline for travel time estimation using large-scale trip data. InProceedings of the ACM SIGSPATIAL International Conference on Advances in Geographic Information Systems. ACM, 61

2016
[33]

Yilun Wang, Yu Zheng, and Yexiang Xue. 2014. Travel time estimation of a path using sparse trajectories. In Proceedings of the ACM SIGKDD international conference on Knowledge discovery and data mining . ACM, 25–34

2014
[34]

Zheng Wang, Kun Fu, and Jieping Ye. 2018. Learning to estimate the travel time. In Proceedings of the ACM SIGKDD international conference on Knowledge discovery and data mining . ACM, 858–866

2018
[35]

Bin Yang, Jian Dai, Chenjuan Guo, Christian S Jensen, and Jilin Hu. 2018. PACE: a PAth-CEntric paradigm for stochastic path finding. The International Journal on Very Large Data Bases 27, 2 (2018), 153–178

2018
[36]

Bin Yang, Chenjuan Guo, and Christian S Jensen. 2013. Travel cost inference from sparse, spatio temporally correlated time series using Markov models.Proceedings of the VLDB Endowment 6, 9 (2013), 769–780

2013
[37]

Jing Yuan, Yu Zheng, Xing Xie, and Guangzhong Sun. 2013. T-Drive: Enhanc- ing Driving Directions with Taxi Drivers’ Intelligence. IEEE Transactions on Knowledge Data Engineering 25, 1 (2013), 220–232

2013
[38]

Hanyuan Zhang, Hao Wu, Weiwei Sun, and Baihua Zheng. 2018. DeepTravel: a Neural Network Travel Time Estimation with Auxiliary Supervision. (2018)

2018

Pith tools

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

[1] [20]

Xiaolei Ma, Zhimin Tao, Yinhai Wang, Haiyang Yu, and Yunpeng Wang. 2015. Long short-term memory neural network for traffic speed prediction using remote microwave sensor data. Transportation Research Part C: Emerging Technologies 54 (2015), 187–197

2015

[2] [1]

OpenStreetMap

2004. OpenStreetMap. https://www.openstreetmap.org/

work page 2004

[3] [2]

Taxi Service Trajectory (TST) Prediction Challenge

2015. Taxi Service Trajectory (TST) Prediction Challenge. http://www.geolink. pt/ecmlpkdd2015-challenge/

work page 2015

[4] [3]

Taxi Travel Time Prediction Challenge

2016. Taxi Travel Time Prediction Challenge. http://www.dcjingsai.com

work page 2016

[5] [4]

Muhammad Tayyab Asif, Justin Dauwels, Chong Yang Goh, Ali Oran, Esmail Fathi, Muye Xu, Menoth Mohan Dhanya, Nikola Mitrovic, and Patrick Jaillet. 2014. Spatiotemporal patterns in large-scale traffic speed prediction. IEEE Transactions on Intelligent Transportation Systems 15, 2 (2014), 794–804

work page 2014

[6] [5]

Corrado De Fabritiis, Roberto Ragona, and Gaetano Valenti. 2008. Traffic esti- mation and prediction based on real time floating car data. In International IEEE Conference on Intelligent Transportation Systems . IEEE, 197–203

work page 2008

[7] [6]

Robert Geirhos, Patricia Rubisch, Claudio Michaelis, Matthias Bethge, Felix A Wichmann, and Wieland Brendel. 2018. ImageNet-trained CNNs are biased towards texture; increasing shape bias improves accuracy and robustness. arXiv preprint arXiv:1811.12231 (2018)

arXiv 2018

[8] [7]

Klaus Greff, Rupesh K Srivastava, Jan Koutník, Bas R Steunebrink, and Jürgen Schmidhuber. 2017. LSTM: A search space odyssey. IEEE transactions on neural networks and learning systems 28, 10 (2017), 2222–2232

2017