Тренування нейронної мережі для прогнозування попиту на пасажирські перевезення таксі за допомогою графічних процесорів

dc.citation.epage36
dc.citation.issue1
dc.citation.journalTitleУкраїнський журнал інформаційних технологій
dc.citation.spage29
dc.citation.volume2
dc.contributor.affiliationНаціональний університет “Львівська політехніка”
dc.contributor.affiliationLviv Polytechnic National University
dc.contributor.authorЗгоба, М. І.
dc.contributor.authorГрицюк, Юрій Іванович
dc.contributor.authorZghoba, M. I.
dc.contributor.authorHrytsiuk, Yu. I.
dc.coverage.placenameЛьвів
dc.coverage.placenameLviv
dc.date.accessioned2022-05-24T11:10:10Z
dc.date.available2022-05-24T11:10:10Z
dc.date.created2020-09-23
dc.date.issued2020-09-23
dc.description.abstractРозглянуто особливості тренування нейронної мережі для прогнозування попиту на пасажирські перевезення таксі за допомогою графічних процесорів, що дало змогу пришвидшити процедуру навчання за різних наборів вхідних даних і конфігурацій апаратного забезпечення та його потужності. З'ясовано, що послуги таксі стають доступнішими для більшої кількості людей. Найважливішим завданням будь-якої компанії та водія таксі є мінімізація тривалості очікування нових замовлень та відстані до клієнтів на момент їх замовлення. Аби досягти цієї мети, потрібно мати розуміння транспортної логістики та вміння оцінити географічний попит на перевезення залежно від багатьох чинників. Розглянуто приклад тренування нейронної мережі для передбачення попиту на пасажирські перевезення таксі. Встановлено, щоб нейронна мережа давала хороші прогнози, необхідно обробити великий набір вхідних даних. Оскільки навчання нейронної мережі – це довготривалий процес, то для вирішення цієї проблеми було застосовано розпаралелювання процедури навчання мережі з використанням графічних процесорів. Проведено навчання нейронної мережі на центральному процесорі, одному та двох графічних процесорах відповідно, виконано порівняння тривалості процедури навчання мережі для однієї епохи. Оцінено вплив кількості використаних графічних процесорів на тривалість тренування нейронної мережі у двох різних конфігураціях апаратного забезпечення та його потужності. Тренування мережі здійснено за допомогою набору даних, який містить 4.5 млн поїздок у межах одного міста. Результати дослідження показують, що пришвидшення процедури навчання за допомогою графічних процесорів не завжди дає позитивний результат, позаяк залежить від багатьох чинників – розміру вибірки вхідних даних, правильного поділу вибірки даних на менші підвибірки, а також характеристик апаратного забезпечення та його потужності.
dc.description.abstractThe peculiarities of neural network training for forecasting taxi passenger demand using graphics processing units are considered, which allowed to speed up the training procedure for different sets of input data, hardware configurations, and its power. It has been found that taxi services are becoming more accessible to a wide range of people. The most important task for any transportation company and taxi driver is to minimize the waiting time for new orders and to minimize the distance from drivers to passengers on order receiving. Understanding and assessing the geographical passenger demand that depends on many factors is crucial to achieve this goal. This paper describes an example of neural network training for predicting taxi passenger demand. It shows the importance of a large input dataset for the accuracy of the neural network. Since the training of a neural network is a lengthy process, parallel training was used to speed up the training. The neural network for forecasting taxi passenger demand was trained using different hardware configurations, such as one CPU, one GPU, and two GPUs. The training times of one epoch were compared along with these configurations. The impact of different hardware configurations on training time was analyzed in this work. The network was trained using a dataset containing 4.5 million trips within one city. The results of this study show that the training with GPU accelerators doesn't necessarily improve the training time. The training time depends on many factors, such as input dataset size, splitting of the entire dataset into smaller subsets, as well as hardware and power characteristics.
dc.format.extent29-36
dc.format.pages8
dc.identifier.citationЗгоба М. І. Тренування нейронної мережі для прогнозування попиту на пасажирські перевезення таксі за допомогою графічних процесорів / М. І. Згоба, Ю. І. Грицюк // Український журнал інформаційних технологій. — Львів : Видавництво Львівської політехніки, 2020. — Том 2. — № 1. — С. 29–36.
dc.identifier.citationenZghoba M. I. Training neural network for taxi passenger demand forecasting using graphics processing units / M. I. Zghoba, Yu. I. Hrytsiuk // Ukrainian Journal of Information Technology. — Lviv : Vydavnytstvo Lvivskoi politekhniky, 2020. — Vol 2. — No 1. — P. 29–36.
dc.identifier.urihttps://ena.lpnu.ua/handle/ntb/56899
dc.language.isouk
dc.publisherВидавництво Львівської політехніки
dc.relation.ispartofУкраїнський журнал інформаційних технологій, 1 (2), 2020
dc.relation.ispartofUkrainian Journal of Information Technology, 1 (2), 2020
dc.relation.references[1] Biao Leng, Heng Du, Jianyuan Wang, Li Li, & Zhang Xiong. (2016). Analysis of Taxi Drivers Behaviors Within a Battle Between Two Taxi Apps. IEEE Transactions on Intelligent Transportation Systems, 17(1), 296–300. https://doi.org/10.1109/TITS.2015.2461000
dc.relation.references[2] Bruce Schaller. (2005). A regression model of the number of taxicabs in US cities. Journal of Public Transportation, 8(5), 4–11. http://doi.org/10.5038/2375-0901.8.5.4
dc.relation.references[3] Dhiraj, K. (2019). 10 reasons why PyTorch is the deep learning framework of the future. Retrieved from: https://heartbeat.fritz.ai/10-reasons-why-pytorch-is-the-deep-learning-framework-of-future-6788bd6b5cc2
dc.relation.references[4] Dipanjan Sarkar, Raghav Bali, & Tushar Sharma. (2018). Practical Machine Learning with Python. Springer Science+ Business Media. New York.
dc.relation.references[5] Du, K.-L., & Swamy, M.N.s. (2014). Multilayer Perceptrons: Architecture and Error Backpropagation. Neural Networks and Statistical Learning, pp. 83–126. https://doi.org/10.1007/978-1-4471-5571-3_4
dc.relation.references[6] Fei Miao, Shuo Han, Shan Lin, Qian Wang, John A. Stankovic, Abdeltawab Hendawi, Desheng Zhang, Tain He, & George J. Pappas. (2019). Data-Driven Robust Taxi Dispatch Under Demand Uncertainties. IEEE Transactions on Control Systems Technology, 17(1), 175–191. https://doi.org/10.1109/TCST.2017.2766042
dc.relation.references[7] Firmino, P., de Mattos, Neto P., & Ferreira, T. (2014). Correcting and combining time series forecasters. Neural Networks,50, 1–11.
dc.relation.references[8] Girshick, R., Donahue, J., Darrell, T., & Malik, J. (2016). Region- Based Convolutional Networks for Accurate Object Detection and Segmentation. In: IEEE Transactions on Pattern Analysis and Machine Intelligence, 38(1), 142–158. https://doi.org/10.1109/TPAMI.2015.2437384
dc.relation.references[9] Grossberg, S. Z. (2010). Neural Networks and Natural Intelligence. Cambridge, MA: MIT Press, 651 p.
dc.relation.references[10] Haykin, S. (2008). Neural Networks and Learning Machines. New Jersey: Prentice Hall, 936 p.
dc.relation.references[11] Jason Dsouza. (2020). What is a GPU and do you need one in Deep Learning? Retrieved from: https://towardsdatascience.com/what-is-a-gpu-and-do-you-need-one-in-deep-learning-718b9597aa0d
dc.relation.references[12] John Grinberg, Arzav Jain, & Arzav Vivek (2014). Predicting Taxi Pickups in New York City. Retrieved from: http://robots.stanford.edu/cs221/2016/restricted/projects/vhchoksi/final.pdf.
dc.relation.references[13] Jun Xu, Rouhollah Rahmatizadeh, Ladislau Bölöni, & Damla Turgut. (2018). Real-Time Prediction of Taxi Demand Using Recurrent Neural Networks. IEEE Transaction on Intelligent transport system, 19(8), 2572–2581. https://doi.org/10.1109/TITS.2017.2755684
dc.relation.references[14] Kennedy, R. K., Khoshgoftaar, T. M., Villanustre, F., & Humphrey, T. (2019). A parallel and distributed stochastic gradient descent implementation using commodity clusters. Journal of Big Data, 6(1), 16. https://doi.org/10.1186/s40537-019-0179-2
dc.relation.references[15] Kiani, K. (2005). Detecting business cycle asymmetries using artificial neural networks and time series models. Computational Economics, 26(1), 65–89.
dc.relation.references[16] Kim, Yoon. (2014). Convolutional neural networks for sentence classification. IEMNLP, 1746–1751.
dc.relation.references[17] Kingma, D. P., & Ba, J. (2014). Adam: A method for stochastic optimization. arXiv – preprint arXiv: 1412.6980.
dc.relation.references[18] Krizhevsky Alex, Sutskever Ilya, Hinton Geoffrey E. (2012). Imagenet classification with deep convolutional neural networks. NIPS, 1106–1114.
dc.relation.references[19] Krizhevsky, A. (2014). One weird trick for parallelizing convolutional neural networks. CoRR, abs/1404.5997.
dc.relation.references[20] Lam, M. (2004). Neural network techniques for financial performance prediction: integrating fundamental and technical analysis. Decision Support Systems, 37(4), 567–581.
dc.relation.references[21] Li, J., Nicolae, B., Wozniak, J., & Bosilca, G. (2019). Understanding scalability and fine-grain parallelism of synchronous data parallel training. IEEE/ACM Workshop – Machine Learning in High Performance Computing Environments (MLHPC) IEEE, pp. 1–8. https://doi.org/10.1109/MLHPC49564.2019.00006
dc.relation.references[22] Lopatko, O., & Mykytyn, I. (2016). Neural networks as the means of forecasting the temperature value of a transient process. Measuring Equipment and Metrology, 77, 65–69.
dc.relation.references[23] Luis Moreira-Matias, et al. (2012). A predictive model for the passenger demand on a taxi network. International IEEE Conference on. IEEE, 15, 1014–1019. https://doi.org/10.1109/ITSC.2012.6338680
dc.relation.references[24] Naoto Mukai, & Naoto Yoden. (2012). Taxi Demand Forecasting Based on Taxi Probe Data by Neural Network. Intelligent Interactive Multimedia: Systems and Services. Ed. by Toyohide Watanabe et al. Smart Innovation, Systems and Technologies 14. Springer Berlin Heidelberg, pp. 589–597. https://doi.org/10.1007/978-3-642-29934-6_57
dc.relation.references[25] Nicholas Jing Yuan, Yu Zheng, Liuhang Zhang, & Xing Xie. (2013). T-Finder: A Recommender System for Finding Passengers and Vacant Taxis. IEEE Transactions on Knowledge and Data Engineering, 25(10), 2390–2403. https://doi.org/10.1109/TKDE.2012.153
dc.relation.references[26] Önder, E., Fɪrat, B., & Hepsen, A. (2013). Forecasting Macroeconomic Variables using Artificial Neural Network and Traditional Smoothing Techniques. Journal of Applied Finance & Banking, 3(4), 73–104.
dc.relation.references[27] Pal, S., Ebrahimi, E., Zulfiqar, A., Fu, Y., Zhang, V., Migacz, S., Nellans, D., & Gupta, P. (2019). Optimizing multi-gpu parallelization strategies for deep learning training. EEE Micro, 39(5), 91–101. https://doi.org/10.1109/MM.2019.2935967
dc.relation.references[28] PyTorch. (2020). PyTorch documentation. Retrieved from: https://pytorch.org/docs/stable/index.html
dc.relation.references[29] Ren, S., He, K., Girshick, R., & Sun, J. (2017). Faster RCNN: Towards Real-Time Object Detection with Region Proposal Networks. In: IEEE Transactions on Pattern Analysis and Machine Intelligence, 39(6), 1137–1149. https://doi.org/10.1109/TPAMI.2016.2577031
dc.relation.references[30] Simonyan, K., & Zisserman, A. (2014). Very Deep Convolutional Networks for Large–Scale Image Recognition. CoRR, abs/1409.1556. https://doi.org/10.1.1.740.6937
dc.relation.references[31] YouTube. (2020). Consumer assessment of taxi services in large cities. Retrieved from: https://www.youtube.com/watch?v=RE2j1B7EdQM. [In Ukrainian].
dc.relation.references[32] Zhang Xiang, Zhao Junbo, LeCun Yann. (2015). Characterlevel convolutional networks for text classification. In: Advances in Neural Information Processing Systems, pp. 649–657.
dc.relation.referencesen[1] Biao Leng, Heng Du, Jianyuan Wang, Li Li, & Zhang Xiong. (2016). Analysis of Taxi Drivers Behaviors Within a Battle Between Two Taxi Apps. IEEE Transactions on Intelligent Transportation Systems, 17(1), 296–300. https://doi.org/10.1109/TITS.2015.2461000
dc.relation.referencesen[2] Bruce Schaller. (2005). A regression model of the number of taxicabs in US cities. Journal of Public Transportation, 8(5), 4–11. http://doi.org/10.5038/2375-0901.8.5.4
dc.relation.referencesen[3] Dhiraj, K. (2019). 10 reasons why PyTorch is the deep learning framework of the future. Retrieved from: https://heartbeat.fritz.ai/10-reasons-why-pytorch-is-the-deep-learning-framework-of-future-6788bd6b5cc2
dc.relation.referencesen[4] Dipanjan Sarkar, Raghav Bali, & Tushar Sharma. (2018). Practical Machine Learning with Python. Springer Science+ Business Media. New York.
dc.relation.referencesen[5] Du, K.-L., & Swamy, M.N.s. (2014). Multilayer Perceptrons: Architecture and Error Backpropagation. Neural Networks and Statistical Learning, pp. 83–126. https://doi.org/10.1007/978-1-4471-5571-3_4
dc.relation.referencesen[6] Fei Miao, Shuo Han, Shan Lin, Qian Wang, John A. Stankovic, Abdeltawab Hendawi, Desheng Zhang, Tain He, & George J. Pappas. (2019). Data-Driven Robust Taxi Dispatch Under Demand Uncertainties. IEEE Transactions on Control Systems Technology, 17(1), 175–191. https://doi.org/10.1109/TCST.2017.2766042
dc.relation.referencesen[7] Firmino, P., de Mattos, Neto P., & Ferreira, T. (2014). Correcting and combining time series forecasters. Neural Networks,50, 1–11.
dc.relation.referencesen[8] Girshick, R., Donahue, J., Darrell, T., & Malik, J. (2016). Region- Based Convolutional Networks for Accurate Object Detection and Segmentation. In: IEEE Transactions on Pattern Analysis and Machine Intelligence, 38(1), 142–158. https://doi.org/10.1109/TPAMI.2015.2437384
dc.relation.referencesen[9] Grossberg, S. Z. (2010). Neural Networks and Natural Intelligence. Cambridge, MA: MIT Press, 651 p.
dc.relation.referencesen[10] Haykin, S. (2008). Neural Networks and Learning Machines. New Jersey: Prentice Hall, 936 p.
dc.relation.referencesen[11] Jason Dsouza. (2020). What is a GPU and do you need one in Deep Learning? Retrieved from: https://towardsdatascience.com/what-is-a-gpu-and-do-you-need-one-in-deep-learning-718b9597aa0d
dc.relation.referencesen[12] John Grinberg, Arzav Jain, & Arzav Vivek (2014). Predicting Taxi Pickups in New York City. Retrieved from: http://robots.stanford.edu/cs221/2016/restricted/projects/vhchoksi/final.pdf.
dc.relation.referencesen[13] Jun Xu, Rouhollah Rahmatizadeh, Ladislau Bölöni, & Damla Turgut. (2018). Real-Time Prediction of Taxi Demand Using Recurrent Neural Networks. IEEE Transaction on Intelligent transport system, 19(8), 2572–2581. https://doi.org/10.1109/TITS.2017.2755684
dc.relation.referencesen[14] Kennedy, R. K., Khoshgoftaar, T. M., Villanustre, F., & Humphrey, T. (2019). A parallel and distributed stochastic gradient descent implementation using commodity clusters. Journal of Big Data, 6(1), 16. https://doi.org/10.1186/s40537-019-0179-2
dc.relation.referencesen[15] Kiani, K. (2005). Detecting business cycle asymmetries using artificial neural networks and time series models. Computational Economics, 26(1), 65–89.
dc.relation.referencesen[16] Kim, Yoon. (2014). Convolutional neural networks for sentence classification. IEMNLP, 1746–1751.
dc.relation.referencesen[17] Kingma, D. P., & Ba, J. (2014). Adam: A method for stochastic optimization. arXiv – preprint arXiv: 1412.6980.
dc.relation.referencesen[18] Krizhevsky Alex, Sutskever Ilya, Hinton Geoffrey E. (2012). Imagenet classification with deep convolutional neural networks. NIPS, 1106–1114.
dc.relation.referencesen[19] Krizhevsky, A. (2014). One weird trick for parallelizing convolutional neural networks. CoRR, abs/1404.5997.
dc.relation.referencesen[20] Lam, M. (2004). Neural network techniques for financial performance prediction: integrating fundamental and technical analysis. Decision Support Systems, 37(4), 567–581.
dc.relation.referencesen[21] Li, J., Nicolae, B., Wozniak, J., & Bosilca, G. (2019). Understanding scalability and fine-grain parallelism of synchronous data parallel training. IEEE/ACM Workshop – Machine Learning in High Performance Computing Environments (MLHPC) IEEE, pp. 1–8. https://doi.org/10.1109/MLHPC49564.2019.00006
dc.relation.referencesen[22] Lopatko, O., & Mykytyn, I. (2016). Neural networks as the means of forecasting the temperature value of a transient process. Measuring Equipment and Metrology, 77, 65–69.
dc.relation.referencesen[23] Luis Moreira-Matias, et al. (2012). A predictive model for the passenger demand on a taxi network. International IEEE Conference on. IEEE, 15, 1014–1019. https://doi.org/10.1109/ITSC.2012.6338680
dc.relation.referencesen[24] Naoto Mukai, & Naoto Yoden. (2012). Taxi Demand Forecasting Based on Taxi Probe Data by Neural Network. Intelligent Interactive Multimedia: Systems and Services. Ed. by Toyohide Watanabe et al. Smart Innovation, Systems and Technologies 14. Springer Berlin Heidelberg, pp. 589–597. https://doi.org/10.1007/978-3-642-29934-6_57
dc.relation.referencesen[25] Nicholas Jing Yuan, Yu Zheng, Liuhang Zhang, & Xing Xie. (2013). T-Finder: A Recommender System for Finding Passengers and Vacant Taxis. IEEE Transactions on Knowledge and Data Engineering, 25(10), 2390–2403. https://doi.org/10.1109/TKDE.2012.153
dc.relation.referencesen[26] Önder, E., Fɪrat, B., & Hepsen, A. (2013). Forecasting Macroeconomic Variables using Artificial Neural Network and Traditional Smoothing Techniques. Journal of Applied Finance & Banking, 3(4), 73–104.
dc.relation.referencesen[27] Pal, S., Ebrahimi, E., Zulfiqar, A., Fu, Y., Zhang, V., Migacz, S., Nellans, D., & Gupta, P. (2019). Optimizing multi-gpu parallelization strategies for deep learning training. EEE Micro, 39(5), 91–101. https://doi.org/10.1109/MM.2019.2935967
dc.relation.referencesen[28] PyTorch. (2020). PyTorch documentation. Retrieved from: https://pytorch.org/docs/stable/index.html
dc.relation.referencesen[29] Ren, S., He, K., Girshick, R., & Sun, J. (2017). Faster RCNN: Towards Real-Time Object Detection with Region Proposal Networks. In: IEEE Transactions on Pattern Analysis and Machine Intelligence, 39(6), 1137–1149. https://doi.org/10.1109/TPAMI.2016.2577031
dc.relation.referencesen[30] Simonyan, K., & Zisserman, A. (2014). Very Deep Convolutional Networks for Large–Scale Image Recognition. CoRR, abs/1409.1556. https://doi.org/10.1.1.740.6937
dc.relation.referencesen[31] YouTube. (2020). Consumer assessment of taxi services in large cities. Retrieved from: https://www.youtube.com/watch?v=RE2j1B7EdQM. [In Ukrainian].
dc.relation.referencesen[32] Zhang Xiang, Zhao Junbo, LeCun Yann. (2015). Characterlevel convolutional networks for text classification. In: Advances in Neural Information Processing Systems, pp. 649–657.
dc.relation.urihttps://doi.org/10.1109/TITS.2015.2461000
dc.relation.urihttp://doi.org/10.5038/2375-0901.8.5.4
dc.relation.urihttps://heartbeat.fritz.ai/10-reasons-why-pytorch-is-the-deep-learning-framework-of-future-6788bd6b5cc2
dc.relation.urihttps://doi.org/10.1007/978-1-4471-5571-3_4
dc.relation.urihttps://doi.org/10.1109/TCST.2017.2766042
dc.relation.urihttps://doi.org/10.1109/TPAMI.2015.2437384
dc.relation.urihttps://towardsdatascience.com/what-is-a-gpu-and-do-you-need-one-in-deep-learning-718b9597aa0d
dc.relation.urihttp://robots.stanford.edu/cs221/2016/restricted/projects/vhchoksi/final.pdf
dc.relation.urihttps://doi.org/10.1109/TITS.2017.2755684
dc.relation.urihttps://doi.org/10.1186/s40537-019-0179-2
dc.relation.urihttps://doi.org/10.1109/MLHPC49564.2019.00006
dc.relation.urihttps://doi.org/10.1109/ITSC.2012.6338680
dc.relation.urihttps://doi.org/10.1007/978-3-642-29934-6_57
dc.relation.urihttps://doi.org/10.1109/TKDE.2012.153
dc.relation.urihttps://doi.org/10.1109/MM.2019.2935967
dc.relation.urihttps://pytorch.org/docs/stable/index.html
dc.relation.urihttps://doi.org/10.1109/TPAMI.2016.2577031
dc.relation.urihttps://doi.org/10.1.1.740.6937
dc.relation.urihttps://www.youtube.com/watch?v=RE2j1B7EdQM
dc.rights.holder© Національний університет “Львівська політехніка”, 2020
dc.subjectмашинне навчання
dc.subjectпрогнозування попиту
dc.subjectтренування нейронної мережі
dc.subjectпришвидшення процедури навчання
dc.subjectпаралелізація процедури тренування
dc.subjectmachine learning
dc.subjectdemand forecasting
dc.subjectneural network training
dc.subjecttraining speedup
dc.subjecttraining parallelization
dc.titleТренування нейронної мережі для прогнозування попиту на пасажирські перевезення таксі за допомогою графічних процесорів
dc.title.alternativeTraining neural network for taxi passenger demand forecasting using graphics processing units
dc.typeArticle

Files

Original bundle

Now showing 1 - 2 of 2
Thumbnail Image
Name:
2020v2n1_Zghoba_M_I-Training_neural_network_29-36.pdf
Size:
1.94 MB
Format:
Adobe Portable Document Format
Thumbnail Image
Name:
2020v2n1_Zghoba_M_I-Training_neural_network_29-36__COVER.png
Size:
1.81 MB
Format:
Portable Network Graphics

License bundle

Now showing 1 - 1 of 1
No Thumbnail Available
Name:
license.txt
Size:
1.82 KB
Format:
Plain Text
Description: