一种改进的深度置信网络在交通流预测中的应用
|
摘要
针对现有交通流预测方法忽视对交通流数据自身特征的有效利用以及不能模拟更复杂的数学运算,提出了一种改进深度置信网络(deep belief network,DBN)的交通流预测方法。该方法结合深度置信网络模型与softmax回归作为预测模型,利用连续受限玻尔兹曼机(continuous restricted Boltzmann machines,CRBM)处理输入特征向量,利用自适应学习步长(adaptive learning step,ALS)减少RBM训练网络模型时重建误差所需的时间,用改进的深度置信网络模型进行交通流特征学习,在网络顶层连接softmax回归模型进行流量预测。实验结果表明,在实际的交通流数据预测中,改进的DBN模型的预测准确率以及时间复杂度相比传统预测模型都得到了较好的改善。
In view of the fact that the existing traffic flow prediction methods ignore the efficient use of the traffic flow data characteristics and can not simulate more complex mathematical operations,this paper proposed a traffic flow prediction method to improve the deep belief network( DBN). The proposed method combined deep belief network model and softmax regression as prediction models,and used the continuous restricted Boltzmann machine( CRBM) to process the input eigenvectors and reduced the adaptive learning step( ALS) RBM trains the time needed to reconstruct the network model,and used the improved deep belief network model to study the traffic flow characteristics. It connected softmax regression model to the top of the network for traffic prediction. The experimental results show that in the actual traffic flow data prediction,the improved prediction accuracy and time complexity of the improved DBN model are better than the traditional prediction model.
In view of the fact that the existing traffic flow prediction methods ignore the efficient use of the traffic flow data characteristics and can not simulate more complex mathematical operations,this paper proposed a traffic flow prediction method to improve the deep belief network( DBN). The proposed method combined deep belief network model and softmax regression as prediction models,and used the continuous restricted Boltzmann machine( CRBM) to process the input eigenvectors and reduced the adaptive learning step( ALS) RBM trains the time needed to reconstruct the network model,and used the improved deep belief network model to study the traffic flow characteristics. It connected softmax regression model to the top of the network for traffic prediction. The experimental results show that in the actual traffic flow data prediction,the improved prediction accuracy and time complexity of the improved DBN model are better than the traditional prediction model.
引文
[1]江德浩.基于深度信念网络的短时交通流预测[D].长沙:湖南师范大学,2016.(Jiang Dehao. Short-term traffic flow predictionbased on deep belief network[D]. Changsha:Hunan Normal Univer-sity,2016.)
[2]刘建伟,刘媛,罗雄麟.深度学习研究进展[J].计算机应用研究,2014,31(7):1921-1930,1942.(Liu Jianwei,Liu Yuan,Luo Xiong-lin. Progress in deep learning[J]. Application Research of Com-puters,2014,31(7):1921-1930,1942.)
[3]罗向龙,焦琴琴,牛力瑶,等.基于深度学习的短时交通流预测[J].计算机应用研究,2017,34(1):91-93,97.(Luo Xianglong,Jiao Qinqin,Niu Liyao,et al. Short-term traffic flow prediction basedon deep learning[J]. Application Research of Computers,2017,34(1):91-93,97.)
[4] Hinton G E,Osindero S,Teh Y W. A fast learning algorithm for deepbelief nets[J]. Neural Computation,2006,18(7):1527-1554.
[5] Fischer A,Igel C. Training restricted Boltzmann machines:an intro-duction[J]. Pattern Recognition,2014,47(1):25-39.
[6] Hinton G E. A practical guide to training restricted Boltzmann ma-chines[J]. Momentum,2012,9(1):599-619.
[7] Salakhutdinov R,Hinton G. Using deep belief nets to learn covariancekernels for Gaussian processes[C]//Proc of International Conferenceon Neural Information Processing Systems.[S. l.]:Curran AssociatesInc.,2007:1249-1256.
[8] Chen H,Murray A F. Continuous restricted Boltzmann machine withan implementable training algorithm[J]. IEE Proceedings:Vision,Image and Signal Processing,2003,150(3):153-158.
[9]柴瑞敏,佀称称.一种改进的深度置信网络及其在自然图像分类中的应用[J].计算机应用与软件,2016,33(9):221-223.(ChaiRuimin,Si Chenchen. An improved deep belief network and its appli-cation in natural image classification[J]. Computer Applicationsand Software,2016,33(9):221-223.)
[10]Bengio Y,Lamblin P,Popovici D,et al. Greedy layer-wise training ofdeep networks[C]//Advances in Neural Information Processing Sys-tems. 2007:153-160.
[11]Silva F M,Almeida L B. Acceleration techniques for the backpropaga-tion algorithm[C]//Proc of Eurasip Workshop on Neural Networks.Berlin:Springer-Verlag,1990:110-119.
[12]刘思妍.城市道路交通流预测方法的研究[D].上海:上海交通大学,2015.(Liu Siyan. Research on forecast method of urban roadtraffic flow[D]. Shanghai:Shanghai Jiaotong University,2015.)
[13]尹邵龙,赵亚楠.深度学习在城市交通流预测中的实践研究[J].现代电子技术,2015,38(15):158-162.(Yin Shaolong,ZhaoYanan. Practical research on deep learning in urban traffic flow fore-casting[J]. Modern Electronics Technique,2015,38(15):158-162.)
[14]孙志军,薛磊,许阳明,等.深度学习研究综述[J].计算机应用研究,2012,29(8):2806-2810.(Sun Zhijun,Xue Lei,Xu Yangming,et al. Overview of deep learning[J]. Application Research ofComputers,2012,29(8):2806-2810.)
[15] Lyu Yisheng,Duan Yanjie,Kang Wenwen,et al. Traffic flow predic-tion with big data:a deep learning approach[J]. IEEE Trans on In-telligent Transportation Systems,2015,16(2):865-873.
[2]刘建伟,刘媛,罗雄麟.深度学习研究进展[J].计算机应用研究,2014,31(7):1921-1930,1942.(Liu Jianwei,Liu Yuan,Luo Xiong-lin. Progress in deep learning[J]. Application Research of Com-puters,2014,31(7):1921-1930,1942.)
[3]罗向龙,焦琴琴,牛力瑶,等.基于深度学习的短时交通流预测[J].计算机应用研究,2017,34(1):91-93,97.(Luo Xianglong,Jiao Qinqin,Niu Liyao,et al. Short-term traffic flow prediction basedon deep learning[J]. Application Research of Computers,2017,34(1):91-93,97.)
[4] Hinton G E,Osindero S,Teh Y W. A fast learning algorithm for deepbelief nets[J]. Neural Computation,2006,18(7):1527-1554.
[5] Fischer A,Igel C. Training restricted Boltzmann machines:an intro-duction[J]. Pattern Recognition,2014,47(1):25-39.
[6] Hinton G E. A practical guide to training restricted Boltzmann ma-chines[J]. Momentum,2012,9(1):599-619.
[7] Salakhutdinov R,Hinton G. Using deep belief nets to learn covariancekernels for Gaussian processes[C]//Proc of International Conferenceon Neural Information Processing Systems.[S. l.]:Curran AssociatesInc.,2007:1249-1256.
[8] Chen H,Murray A F. Continuous restricted Boltzmann machine withan implementable training algorithm[J]. IEE Proceedings:Vision,Image and Signal Processing,2003,150(3):153-158.
[9]柴瑞敏,佀称称.一种改进的深度置信网络及其在自然图像分类中的应用[J].计算机应用与软件,2016,33(9):221-223.(ChaiRuimin,Si Chenchen. An improved deep belief network and its appli-cation in natural image classification[J]. Computer Applicationsand Software,2016,33(9):221-223.)
[10]Bengio Y,Lamblin P,Popovici D,et al. Greedy layer-wise training ofdeep networks[C]//Advances in Neural Information Processing Sys-tems. 2007:153-160.
[11]Silva F M,Almeida L B. Acceleration techniques for the backpropaga-tion algorithm[C]//Proc of Eurasip Workshop on Neural Networks.Berlin:Springer-Verlag,1990:110-119.
[12]刘思妍.城市道路交通流预测方法的研究[D].上海:上海交通大学,2015.(Liu Siyan. Research on forecast method of urban roadtraffic flow[D]. Shanghai:Shanghai Jiaotong University,2015.)
[13]尹邵龙,赵亚楠.深度学习在城市交通流预测中的实践研究[J].现代电子技术,2015,38(15):158-162.(Yin Shaolong,ZhaoYanan. Practical research on deep learning in urban traffic flow fore-casting[J]. Modern Electronics Technique,2015,38(15):158-162.)
[14]孙志军,薛磊,许阳明,等.深度学习研究综述[J].计算机应用研究,2012,29(8):2806-2810.(Sun Zhijun,Xue Lei,Xu Yangming,et al. Overview of deep learning[J]. Application Research ofComputers,2012,29(8):2806-2810.)
[15] Lyu Yisheng,Duan Yanjie,Kang Wenwen,et al. Traffic flow predic-tion with big data:a deep learning approach[J]. IEEE Trans on In-telligent Transportation Systems,2015,16(2):865-873.