AdaBelief Optimizasyon Tekniğinin Derin Öğrenmeye Dayalı Yaya Rotası Tahmin Uygulamalarına Etkisinin “Yakınsama” açısından İncelenmesi

Yıl 2024, Cilt: 6 Sayı: 1, 1 - 10, 15.06.2024

Sevcan Turan , Feyzullah Temurtaş

https://doi.org/10.55213/kmujens.1418280

Öz

Son yıllarda, görüntü işleme teknikleri kullanılarak yayaların takip edebileceği rotanın tahmini, hızla dikkat çeken bir araştırma konusu haline gelmiştir. Rota tahmin uygulamalarında Derin Öğrenmenin kullanımı, mühendislik çalışmalarıyla yapılan geleneksel parametre belirleme işlemlerine ihtiyaç duymayan ve daha doğru tahminler yapabilen yeni uygulamaların geliştirilmesini sağlamıştır. Rota tahmini için sıklıkla veriye dayalı olan gözetimli derin öğrenme modelleri kullanılmaktadır. Ancak, bu modellerin eğitimi yüksek hesaplama maliyeti getirmektedir. Bu maliyetleri azaltmak ve doğrulukları arttırmak için iyi yakınsama ve genelleştirme özelliklerine sahip optimizasyon yöntemlerini seçmek önemlidir. Bu çalışma, ETH/UCY veri kümeleri kullanılarak derin öğrenme mimarisi temelli geliştirilmiş rota tahmini algoritmalarının optimizasyon yöntemi açısından performansını incelemektedir. Özellikle, modelin eğitimi aşamasında yakınsama açısından AdaBelief optimizasyon tekniğinin avantajları ve dezavantajlarına odaklanılmaktadır. Çalışmanın sonuçları, AdaBelief’in eğitim sürecine pozitif bir katkıda bulunduğunu ve rota tahmini algoritmalarının genel performansını arttırabileceğini göstermektedir.

Anahtar Kelimeler

AdaBelief, derin öğrenme, ETH/UCY, yaya rota tahmini, optimizasyon

Investigation of the Effects of AdaBelief Optimization Technique on Deep Learning-Based Pedestrian Path Prediction Applications in terms of “Convergence”

Öz

In recent years, the prediction of pedestrian paths using computer vision techniques has become an increasingly attractive topic of research. The use of deep learning techniques has led to the development of new path prediction applications that do not rely on the traditional parameter determination processes with engineering studies. This has resulted in more accurate predictions. Supervised deep learning models, which are data-driven, have been widely used for path prediction. However, the training of these models is associated with high computational costs. To address this issue, it is important to choose optimization methods that have good convergence and generalization properties in order to reduce costs and improve accuracy. This study examines the performance of path prediction algorithms based on deep learning architectures using the ETH/UCY datasets. In particular, the study focuses on the advantages and disadvantages of the AdaBelief optimization technique in terms of convergence during the training phase. The results of the study show that the AdaBelief makes a positive contribution to the training process and can improve the overall performance of the path prediction algorithm.

Anahtar Kelimeler

AdaBelief, deep learning, ETH/UCY, Pedestrian path prediction, optimization

Kaynakça

• Alahi A, Goel K, Ramanathan V, Robicquet A, Fei-Fei L, Savarese S (2016). Social LSTM: Human Trajectory Prediction in Crowded Spaces. IEEE Conference on Computer Vision and Pattern Recognition. Las Vegas, USA.
• Bera A, Kim S, Randhavane T, Pratapa S, Manocha D (2016). GLMP- realtime pedestrian path prediction using global and local movement patterns. IEEE International Conference on Robotics and Automation. Stockholm, Sweden.
• Bottou L (1991). Stochastic gradient learning in neural networks. Proceedings of Neuro-Nımes, 91(8): 12.
• CARPE (2023). https://github.com/TeCSAR-UNCC/CARPe_Posterum.
• CausalHTP. (2023). https://github.com/CHENGY12/CausalHTP.
• Chen G, Li J, Lu J, Zhou J (2021). Human Trajectory Prediction via Counterfactual Analysis. IEEE/CVF International Conference on Computer Vision. Montreal, Canada.
• Goodfellow I, Pouget-Abadie J, Mirza M, Xu B, Warde-Farley D, Ozair S, Courville A, Bengio Y (2014). Generative adversarial nets. Advances in Neural Information Processing Systems, 27.
• Gulzar M, Muhammad Y, Muhammad N (2021). A Survey on Motion Prediction of Pedestrians and Vehicles for Autonomous Driving. IEEE Access, 9:137957–137969.
• Guo J, Li J, Leng D, Pan L (2021). Heterogeneous Graph based Deep Learning for Biomedical Network Link Prediction. arXiv preprint arXiv:2102.01649.
• Guo S, Fraser MW (2014). Propensity score analysis: Statistical methods and applications (Vol. 11). SAGE Publications.
• Gupta A, Johnson J, Fei-Fei L, Savarese S, Alahi A (2018). Social GAN: Socially Acceptable Trajectories with Generative Adversarial Networks. Salt Lake City, USA.
• Hariyono J, Shahbaz A, Jo K-H (2015). Estimation of walking direction for pedestrian path prediction from moving vehicle. IEEE/SICE International Symposium on System Integration. Nagoya, Japan.
• Hecht J (2018). Lidar for Self-Driving Cars. Optics & Photonics News, 28–33.
• Hochreiter S, Schmidhuber J (1997). Long Short-Term Memory. Neural Computation, 9(8): 1735-1780.
• Huang Y, Bi H, Li Z, Mao T, Wang Z (2019). STGAT: Modeling Spatial-Temporal Interactions for Human Trajectory Prediction. IEEE/CVF International Conference on Computer Vision. Seoul, Korea.
• Jain AB, Casas S, Liao R, Xiong Y, Feng S, Segal S, Urtasun R (2019). Discrete Residual Flow for Probabilistic Pedestrian Behavior Prediction. Conference on Robot Learning. Osaka, Japan.
• Keller CG, Gavrila DM (2014). Will the Pedestrian Cross? A Study on Pedestrian Path Prediction. IEEE Transactions on Intelligent Transportation Systems, 15(2): 494–506.
• Kingma DP, Ba J (2014). Adam: A Method for Stochastic Optimization. arXiv preprint arXiv:1412.6980.
• Kipf TN, Welling M (2016). Semi-Supervised Classification with Graph Convolutional Networks. arXiv preprint arXiv:1609.02907.
• Kolcu C, Bolat B (2018). Yayaların yürüyüş rotalarının belirlenmesi. Electric Electronics, Computer Science, Biomedical Engineerings’ Meeting (EBBT). İstanbul, Türkiye.
• Le Cun Y, Jackel LD, Boser B, Denker JS, Graf HP, Guyon I, Henderson D, Howard RE, Hubbard W (1989). Handwritten digit recognition: applications of neural network chips and automatic learning. IEEE Communications Magazine, 27(11): 41–46.
• Leinonen J (2021). Improvements to short-term weather prediction with recurrent-convolutional networks. IEEE International Conference on Big Data (Big Data). Orlando, FL, USA.
• Lerner A, Chrysanthou Y, Lischinski D (2007). Crowds by Example. Computer Graphics Forum, 26(3): 655–664.
• Liu Y, Zhang M, Zhong Z, Zeng X, Long X (2021). A comparative study of recently deep learning optimizers. International Conference on Algorithms, High Performance Computing, and Artificial Intelligence (AHPCAI 2021). Sanya, China.
• Lv Y, Zhou Q, Li Y, Li W (2021). A predictive maintenance system for multi-granularity faults based on AdaBelief-BP neural network and fuzzy decision making. Advanced Engineering Informatics, 49: 101318.
• Ma Y, Zhu X, Zhang S, Yang R, Wang W, Manocha D (2019). TrafficPredict: Trajectory Prediction for Heterogeneous Traffic-Agents. AAAI Conference on Artificial Intelligence. New Orleans, Louisiana, USA.
• Mendieta, M., & Tabkhi, H. (2021, May). Carpe posterum: A convolutional approach for real-time pedestrian path prediction. AAAI Conference on Artificial Intelligence. Vancouver, Canada.
• Mittal S, Vetter JS (2015). A Survey of Methods for Analyzing and Improving GPU Energy Efficiency. ACM Computing Surveys, 47(2): 1–23.
• Mohamed A, Qian K, Elhoseiny M, Claudel C (2020). Social-stgcnn: A social spatio-temporal graph convolutional neural network for human trajectory prediction. IEEE/CVF Conference on Computer Vision and Pattern Recognition. Seattle, WA, USA.
• Ozyildirim BM, Kiran M (2020). Do optimization methods in deep learning applications matter? arXiv preprint arXiv:2002.12642.
• Pei D, Jing M, Liu H, Sun F, Jiang L (2020). A fast RetinaNet fusion framework for multi-spectral pedestrian detection. Infrared Physics & Technology, 105: 103178.
• Pellegrini S, Ess A, Schindler K, Van Gool L (2009). You’ll never walk alone: Modeling social behavior for multi-target tracking. 2009 IEEE 12th International Conference on Computer Vision. Kyoto, Japan.
• Rudenko A, Palmieri L, Herman M, Kitani KM, Gavrila DM, Arras KO (2019). Human Motion Trajectory Prediction: A Survey. The International Journal of Robotics Research, 39(8): 895-935.
• SGAN (2023). https://github.com/agrimgupta92/sgan.
• SGCN (2023). https://github.com/shuaishiliu/SGCN.
• Shi H, Wang L, Scherer R, Wozniak M. Zhang P, Wei W (2021). Short-Term Load Forecasting Based on Adabelief Optimized Temporal Convolutional Network and Gated Recurrent Unit Hybrid Neural Network. IEEE Access, 9: 66965–66981.
• Shi, L., Wang, L., Long, C., Zhou, S., Zhou, M., Niu, Z., & Hua, G. (2021). SGCN: Sparse graph convolution network for pedestrian trajectory prediction. IEEE/CVF Conference on Computer Vision and Pattern Recognition. Nashville, TN, USA.
• SocialSTGCNN (2023). https://github.com/abduallahmohamed/Social-STGCNN.
• STGAT (2023). https://github.com/huang-xx/STGAT.
• Turan S, Milani B, Temurtaş F (2021). Different Application Areas Of Object Detection With Deep Learning. Akıllı Ulaşım Sistemleri ve Uygulamaları Dergisi. 4(2): 148-164.
• Veličković P, Cucurull G, Casanova A, Romero A, Liò P, Bengio Y (2017). Graph Attention Networks. arXiv preprint arXiv:1710.10903.
• Wang H, Li C, Zhang Y, Liu Z, Hui Y, Mao G (2020). A Scheme on Pedestrian Detection using Multi-Sensor Data Fusion for Smart Roads. 2020 IEEE 91st Vehicular Technology Conference (VTC2020-Spring). Antwerp, Belgium.
• Wang Y, Liu J, Chang X (2021). Generalizing Adversarial Examples by AdaBelief Optimizer. arXiv preprint arXiv:2101.09930.
• Wu Z, Pan S, Chen F, Long G, Zhang C, Yu PS (2019). A Comprehensive Survey on Graph Neural Networks. IEEE Transactions on Neural Networks and Learning Systems, 32(1): 4-24.
• Xue-Wen Chen, Xiaotong Lin (2014). Big Data Deep Learning: Challenges and Perspectives. IEEE Access, 2: 514–525.
• Yang L, Shami A (2020). On hyperparameter optimization of machine learning algorithms: Theory and practice. Neurocomputing, 415: 295–316.
• Zhang S, Tong H, Xu J, Maciejewski R (2019). Graph convolutional networks: a comprehensive review. Computational Social Networks, 6(1): 11.
• Zhou D, Qiu S, Song Y, Xia K (2020). A pedestrian extraction algorithm based on single infrared image. Infrared Physics & Technology, 105: 103236.
• Zhou J, Cui G, Hu S, Zhang Z, Yang C, Liu Z, Wang L, Li C, Sun M (2020). Graph neural networks: A review of methods and applications. AI Open, 1: 57–81.
• Zhu Z, Sun H, Zhang C (2021). Effectiveness of Optimization Algorithms in Deep Image Classification. arXiv preprint arXiv:2110.01598.
• Zhuang J, Tang T, Ding Y, Tatikonda S, Dvornek N, Papademetris X, Duncan JS (2020). AdaBelief Optimizer: Adapting Stepsizes by the Belief in Observed Gradients. Advances in Neural İnformation Processing Systems, 33: 18795-18806.