2020/08/11 by Kazemi, Amir, Hadi Meidani, Meidani, Hadi · 1 citation
Computer Science · Engineering · #Computation (stat.CO) #FOS: Computer and information sciences #Machine Learning (cs.LG) #Machine Learning (stat.ML) #Music and Audio Processing #Traffic Prediction and Management Techniques #Traffic control and management
paper · pdf · doi:10.48550/arxiv.2008.04847
openalex publication_date 2020/08/11 · openalex created_date 2023/03/08 · openalex updated_date 2026/07/28
Increasing use of sensor data in intelligent transportation systems calls for\naccurate imputation algorithms that can enable reliable traffic management in\nthe occasional absence of data. As one of the effective imputation approaches,\ngenerative adversarial networks (GANs) are implicit generative models that can\nbe used for data imputation, which is formulated as an unsupervised learning\nproblem. This work introduces a novel iterative GAN architecture, called\nIterative Generative Adversarial Networks for Imputation (IGANI), for data\nimputation. IGANI imputes data in two steps and maintains the invertibility of\nthe generative imputer, which will be shown to be a sufficient condition for\nthe convergence of the proposed GAN-based imputation. The performance of our\nproposed method is evaluated on (1) the imputation of traffic speed data\ncollected in the city of Guangzhou in China, and the training of short-term\ntraffic prediction models using imputed data, and (2) the imputation of\nmulti-variable traffic data of highways in Portland-Vancouver metropolitan\nregion which includes volume, occupancy, and speed with different missing rates\nfor each of them. It is shown that our proposed algorithm mostly produces more\naccurate results compared to those of previous GAN-based imputation\narchitectures.\n