Reinforcement Learning-designed LSTM for Trajectory and Traffic Flow Prediction

Karimzadeh, Mostafa; Aebi, Ryan; Souza, Allan M. de; Zhao, Zhongliang; Braun, Torsten; Sargento, Susana; Villas, Leandro A.

doi:10.1109/wcnc49053.2021.9417511

Cited by 14 publications

(6 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…where 𝑥 is calculated using equation ( 11): a weighted concatenation operation of the output of the GAT layer 𝐺𝐴𝑇(•), traffic data input 𝑋 , and the other feature matrix 𝐷(𝑡) (also see block 𝑎 in Figure 1). 𝑥 = 𝑎 𝑋 ∥ 𝑎 𝐺𝐴𝑇(𝑋 ) ∥ 𝑎 𝐷(𝑡) (11) where ∥ denotes concatenation. The corresponding weight vectors 𝑎 , 𝑎 , and 𝑎 are learnable and assignable, and for each node 𝑖, they satisfy the following equation:…”

Section: Gat-lstm Modelmentioning

confidence: 99%

See 1 more Smart Citation

Adaptive Graph Attention and Long Short-Term Memory-Based Networks for Traffic Prediction

Zhu,

Boada,

Boada

2024

Mathematics

View full text Add to dashboard Cite

While the increased availability of traffic data is allowing us to better understand urban mobility, research on data-driven and predictive modeling is also providing new methods for improving traffic management and reducing congestion. In this paper, we present a hybrid predictive modeling architecture, namely GAT-LSTM, by incorporating graph attention (GAT) and long short-term memory (LSTM) networks for handling traffic prediction tasks. In this architecture, GAT networks capture the spatial dependencies of the traffic network, LSTM networks capture the temporal correlations, and the Dayfeature component incorporates time and external information (such as day of the week, extreme weather conditions, holidays, etc.). A key attention block is designed to integrate GAT, LSTM, and the Dayfeature components as well as learn and assign weights to these different components within the architecture. This method of integration is proven effective at improving prediction accuracy, as shown by the experimental results obtained with the PeMS08 open dataset, and the proposed model demonstrates state-of-the-art performance in these experiments. Furthermore, the hybrid model demonstrates adaptability to dynamic traffic conditions, different prediction horizons, and various traffic networks.

show abstract

Section: Gat-lstm Modelmentioning

confidence: 99%

“…Due to its competitive performance, the LSTM method is often regarded as a baseline in subsequently proposed approaches. Meanwhile, ever-more enhanced LSTM models [10,11] and hybrid LSTM-based models [12][13][14][15][16][17] have been proposed.…”

Section: Introductionmentioning

confidence: 99%

Adaptive Graph Attention and Long Short-Term Memory-Based Networks for Traffic Prediction

Zhu,

Boada,

Boada

2024

Mathematics

View full text Add to dashboard Cite

show abstract

“…The core idea is to assign different weights to the hidden layer states by reasonably allocating attention to different input information to highlight the influence of important information on the results. The weight assignment calculation of the attention mechanism can be expressed according to Equations ( 16)- (17).…”

Section: Attention Mechanismmentioning

confidence: 99%

“…Therefore, machine learning has also been applied to time-series data processing. Examples include pedestrian and vessel trajectory and traffic flow predictions [15][16][17]. Trajectory clustering is a clustering analysis of historical trajectories [18,19] that combines updated state information to correct prediction results and improve the prediction accuracy.…”

mentioning

confidence: 99%

Short-Term Trajectory Prediction Based on Hyperparametric Optimisation and a Dual Attention Mechanism

et al. 2022

View full text Add to dashboard Cite

Highly accurate trajectory prediction models can achieve route optimisation and save airspace resources, which is a crucial technology and research focus for the new generation of intelligent air traffic control. Aiming at the problems of inadequate extraction of trajectory features and difficulty in overcoming the short-term memory of time series in existing trajectory prediction, a trajectory prediction model based on a convolutional neural network-bidirectional long short-term memory (CNN-BiLSTM) network combined with dual attention and genetic algorithm (GA) optimisation is proposed. First, to autonomously mine the data association between input features and trajectory features as well as highlight the influence of important features, an attention mechanism was added to a conventional CNN architecture to develop a feature attention module. An attention mechanism was introduced at the output of the BiLSTM network to form a temporal attention module to enhance the influence of important historical information, and GA was used to optimise the hyperparameters of the model to achieve the best performance. Finally, a multifaceted comparison with other typical time-series prediction models based on real flight data verifies that the prediction model based on hyperparameter optimisation and a dual attention mechanism has significant advantages in terms of prediction accuracy and applicability.

show abstract

“…The application of DRL methods has importantly been increasing for solving real‐world optimization problems over the past few years. It has proven to be effective in playing games [16], traffic flow predictions [17], and optimization and control of self‐driving cars [18]. Moreover, the use of DRL to solve combinatorial optimization problems is also the subject of study in recent days.…”

Section: Introductionmentioning

confidence: 99%

Graph Transformer with Reinforcement Learning for Vehicle Routing Problem

Fellek

Farid

Gebreyesus

et al. 2023

IEEJ Transactions Elec Engng

View full text Add to dashboard Cite

Vehicle routing problem (VRP) is one of the classic combinatorial optimization problems where an optimal tour to visit customers is required with a minimum total cost in the presence of some constraints. Recently, VRP is being solved with the use of deep reinforcement learning (DRL), with node sets considered (represented) as a graph structure. Existing Transformer based DRL solutions for VRP rely only on node information ignoring the role of information on the edges between nodes in the graph structure. In this paper, we proposed an attention‐based end‐to‐end DRL model to solve VRP which embeds edge information between nodes for rich graph representation learning. We used Transformer based encoder‐decoder framework with an edge information embedded multi‐head attention (EEMHA) layer in the encoder. The EEMHA‐based encoder learns underlying structure of the graph and generates an expressive graph topology representation by merging node and edge information. We trained our model using proximal policy optimization (PPO) with some code‐level optimization techniques. We conducted an experiment on randomly generated instances and on a real‐world data generated from road networks to verify the performance of our proposed model. The result from all experiments show that our model performs better than the existing DRL methods and most of the conventional heuristics with good generalizability from random instance training to real‐world instance testing of different problem size. © 2023 Institute of Electrical Engineers of Japan. Published by Wiley Periodicals LLC.

show abstract

Reinforcement Learning-designed LSTM for Trajectory and Traffic Flow Prediction

Cited by 14 publications

References 11 publications

Adaptive Graph Attention and Long Short-Term Memory-Based Networks for Traffic Prediction

Adaptive Graph Attention and Long Short-Term Memory-Based Networks for Traffic Prediction

Short-Term Trajectory Prediction Based on Hyperparametric Optimisation and a Dual Attention Mechanism

Graph Transformer with Reinforcement Learning for Vehicle Routing Problem

Contact Info

Product

Resources

About