Short-term traffic flow prediction based on spatio-temporal analysis and CNN deep learning

Short term traffic flow prediction of expressway service area based on STL-OMS

Short-term traffic speed prediction is a promising research topic in intelligent transportation systems (ITSs), which also plays an important role in the real-time decision-making of traffic control and guidance systems. However, the urban traffic speed has strong temporal, spatial correlation and the characteristic of complex nonlinearity and randomness, which makes it challenging to accurately and efficiently forecast short-term traffic speeds. We investigate the relevant literature and found that although most methods can achieve good prediction performance with the complete sample data, when there is a certain missing rate in the database, it is difficult to maintain accuracy with these methods. Recent studies have shown that deep learning methods, especially long short-term memory (LSTM) models, have good results in short-term traffic flow prediction. Furthermore, the attention mechanism can properly assign weights to distinguish the importance of traffic time sequences, thereby further improving the computational efficiency of the prediction model. Therefore, we propose a framework for short-term traffic speed prediction, including data preprocessing module and short-term traffic prediction module. In the data preprocessing module, the missing traffic data are repaired to provide a complete dataset for subsequent prediction. In the prediction module, a combined deep learning method that is an attention-based LSTM (ATT-LSTM) model for predicting short-term traffic speed on urban roads is proposed. The proposed framework was applied to the urban road network in Nanshan District, Shenzhen, Guangdong Province, China, with a 30-day traffic speed dataset (floating car data) used as the experimental sample. Results show that the proposed method outperforms other deep learning algorithms (such as recurrent neural network (RNN) and convolutional neural network (CNN)) in terms of both calculating efficiency and prediction accuracy. The attention mechanism can significantly reduce the error of the LSTM model (up to 12.4%) and improves the prediction performance.

Accurate selection of spatial-temporal features is key to the short-term traffic flow prediction model outputting higher quality results, which can effectively reduce the difficulty of constructing the prediction model. The spatial-temporal feature selection of most existing short-term traffic flow prediction models mainly relies on empirical knowledge methods and lacks interpretability. The proposed short-term traffic flow prediction network, named STFP-FGFS, utilizes a filter-genetic feature selection method to better explain the results of short-term traffic flow predictions. It consists of three stages: initial generation of temporal-spatial feature set, filtering, and feature optimization, as well as the predicted model. The initial spatial features are generated based on effective travel time, target time granularity, and vehicle type; that is, original spatial features are replaced by standardized spatial features. Four widely used feature selection methods for short-term traffic flow prediction are applied and compared, evaluating three experimental targets and four types of time granularity using four evaluation indexes. The results show that the STFP-FGFS proposed method has overall superior performance, good interpretability, and readability for selected spatial-temporal features.

Early detection of cutaneous melanoma (CM) is crucial for patient survival, yet avoiding overdiagnosis remains essential. Differentiating CM from benign melanoma simulators (MelSim) is challenging due to overlapping features. Deep learning convolutional neural networks (DL-CNNs) have demonstrated dermatologist-level accuracy in identifying CM. We hypothesized that support from DL-CNN could increase dermatologists' accuracy in differentiating CM from MelSim. This cross-sectional reader study analyzed the diagnostic performance of a DL-CNN and 27 dermatologists for 200 skin lesions (100 CM, 100 MelSim). The DL-CNN assigned malignancy scores ranging from 0 to 1 (> 0.5 indicating malignancy). Dermatologists assessed lesions across three levels: (I) dermoscopy only, (II) full case information (dermoscopy, close-up images, metadata), and (III) full case information plus DL-CNN scores. Primary outcomes were sensitivity, specificity, and ROC-AUC of dermatologists with or without DL-CNN-support (level-II versus -III). The DL-CNN and dermatologists in level-II showed a comparable sensitivity (95 % CI) of 90.0 % (82.6-94.5 %) and 90.1 % (86.9-93.2 %, p = 0.153), respectively. However, the DL-CNN's specificity (67.0 % (57.3-75.4 %) versus 73.2 % (69.1-77.3 %)) and ROC-AUC (0.889 (0.845-0.932) versus 0.951 (0.920-0.982)) were significantly lower than for dermatologists (all p < 0.01). When dermatologists integrated DL-CNN predictions (level-III) their sensitivity increased to 91.4 % (88.3-94.5 %, p < 0.001) without markedly changing specificity (74.2 % (70.6-77.7 %, p = 0.435)) or ROC-AUC (0.954 (0.927-0.982, p = 0.581)). Collaboration with a DL-CNN slightly improved dermatologists' diagnostic accuracy in a mixed CM and MelSim dataset, by increasing sensitivity without a loss of specificity. The DL-CNN's level of sensitivity in this difficult-to-diagnose dataset underlines the potential as an assistant diagnostic tool.

PurposeThe aim of this study was to analyse and review deep learning convolutional neural networks for detecting and diagnosing early-stage dental caries on periapical radiographs.Materials and MethodsIn order to conduct this review, the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines were followed. Studies published from 2015 to 2021 under the keywords (deep convolutional neural network) AND (caries), (deep learning caries) AND (convolutional neural network) AND (caries) were systematically reviewed.ResultsWhen dental caries is improperly diagnosed, the lesion may eventually invade the enamel, dentin, and pulp tissue, leading to loss of tooth function. Rapid and precise detection and diagnosis are vital for implementing appropriate prevention and treatment of dental caries. Radiography and intraoral images are considered to play a vital role in detecting dental caries; nevertheless, studies have shown that 20% of suspicious areas are mistakenly diagnosed as dental caries using this technique; hence, diagnosis via radiography alone without an objective assessment is inaccurate. Identifying caries with a deep convolutional neural network-based detector enables the operator to distinguish changes in the location and morphological features of dental caries lesions. Deep learning algorithms have broader and more profound layers and are continually being developed, remarkably enhancing their precision in detecting and segmenting objects.ConclusionClinical applications of deep learning convolutional neural networks in the dental field have shown significant accuracy in detecting and diagnosing dental caries, and these models hold promise in supporting dental practitioners to improve patient outcomes.

With the urbanization, urban transportation has become a key factor restricting the development of a city. In a big city, it is important to improve the efficiency of urban transportation. The key to realize short-term traffic flow prediction is to learn its complex spatial correlation, temporal correlation and randomness of traffic flow. In this paper, the convolution neural network (CNN) is proposed to deal with spatial correlation among different regions, considering that the large urban areas leads to a relatively deep Network layer. First three gated recurrent unit (GRU) were used to deal with recent time dependence, daily period dependence and weekly period dependence. Considering that each historical period data to forecast the influence degree of the time period is different, three attention mechanism was taken into GRU. Second, a two-layer full connection network was applied to deal with the randomness of short-term flow combined with additional information such as weather data. Besides, the prediction model was established by combining these three modules. Furthermore, in order to verify the influence of spatial correlation on prediction model, an urban functional area identification model was introduced to identify different functional regions. Finally, the proposed model was validated based on the history of New York City taxi order data and reptiles for weather data. The experimental results show that the prediction precision of our model is obviously superior to the mainstream of the existing prediction methods.

This study proposes a short-term traffic flow prediction model that combines community detection-based federated learning with a graph convolutional network (GCN) to alleviate the time-consuming training, higher communication costs, and data privacy risks of global GCNs as the amount of data increases. The federated community GCN (FCGCN) can achieve timely, accurate, and safe traffic state predictions in the era of big traffic data, which is critical for the efficient operation of intelligent transportation systems. The FCGCN prediction process has four steps: dividing the local subnetwork with community detection, local training based on the global parameters, uploading the local model parameters, and constructing a global model prediction based on the aggregated parameters. Numerical results on the PeMS04 and PeMS08 datasets show that the FCGCN outperforms four benchmark models, namely, the long short-term memory (LSTM), convolutional neural network (CNN), ChebNet, and graph attention network (GAT) models. The FCGCN prediction is closer to the real value, with nearly the same performance as the global model at a lower time cost, thus achieving accurate and secure short-term traffic flow predictions with three parameters: flow, speed, and occupancy.

For the purpose of applying information technology to the creation of ink style painting, the algorithm of ink painting rendering based on the deep learning framework and convolutional neural network model is designed and improved. Firstly, the ink style rendering program is written in Python. Secondly, VGG under Caffe architecture and Illustration 2Vec models are transplanted to TensorFlow architecture, and the image is rendered in ink style based on deep learning framework and convolutional neural network model. Finally, based on Node.js, the server-side program for image ink style rendering is built. Among them, Express is adopted as the Web-side framework, and the front-end page effect is completed. The results show that the ink rendering logic program is applicable, and the expected purpose is achieved.

Accurate short-term traffic flow prediction is the foundation of the efficient and proactive management of freeway networks, especially on the abnormal traffic states. The relationship between traffic flow on the current section and the upstream stations can be used for predicting short-term traffic flow. In this paper, we reveal this relationship by the traffic flow structure pattern. The structure pattern can be drawn from real freeway toll data and a few video detective cameras on the freeway segments. Based on the stability pattern, a new traffic flow prediction algorithm has been proposed. Experimental based on real data showed that the prediction method based on structure pattern is an effective approach for traffic flow prediction, especially on the abnormal traffic state.

Timely and accurate traffic flow information is essential for the realization of intelligent transportation systems (ITS). For the existing traffic flow prediction models, only the temporal features were extracted but the spatial features were neglects. A short-term traffic flow prediction model based on Pearson Correlation Coefficient (PCC) and Bi-directional Long Short-Term Memory Network (BiLSTM) model was proposed. The PCC was used to mine the spatial dependence of the road network, and the temporal features of traffic flow were mined using BiLSTM model. The extracted temporal and spatial features were combined to achieve short-term flow prediction. The experimental results show that the PCC-BiLSTM model has higher prediction accuracy and lower error than the prediction models considering only temporal features, which verifies the effectiveness of traffic flow prediction in consideration of both temporal and spatial characteristics.

The existing short-term traffic flow prediction model fails to consider the analysis of the asynchronous temporal and spatial characteristics of the traffic flow, which leads to the prediction accuracy is difficult to meet the requirements of refined traffic control. Thus, a short-term traffic flow prediction method based on the asynchronous temporal and spatial correlation was proposed in this paper. Firstly, based on Pearson-dynamic time warping method, the trend correlation and distance correlation were calculated respectively in predicted link and associated link, then the results of dynamic time warping were standardized, and discrete degree analysis was carried out on the two kinds of correlation value, evaluation index was designed based on variance of road traffic flow correlation, through the SVM method realizing the closed-loop optimization in the selection process of associated links; Secondly, a traffic flow prediction model based on TPD-LSTM was constructed by long and short term memory neural network. Finally, a link of Yusha Road of Xinhua Avenue in Chengdu was selected as the target link for example analysis, the results show that the prediction accuracy of TPD-LSTM improves to 89.62%.

Short-term traffic flow prediction (STFP) is one of the key technologies in Intelligence Transportation System (ITS). With the development of artificial intelligence technology, deep learning has been employed to STFP and certain achievements have been obtained. However, further improvement of STFP accuracy is constrained by the presence of noise, modal aliasing in the modal components and insufficient feature extraction of neural networks. Therefore, to more accurately capture the changing trend of traffic flow in different time periods, and increase the reliability and interpretability of the prediction models, improved variational mode decomposition (IVMD) and self-attention mechanism (SAM) based hybrid convolutional neural network (CNN) and long short term memory network (LSTM) are proposed for STFP in this study, benefited from the modal decomposition of IVMD and redistribution of the neural weights in CNN-LSTM networks by SAM, the proposed IVMD-CNN-LSTM-SAM effectively suppresses the influence of randomness and volatility of traffic flow, and accurately predict the short-term lane occupancy in the designated area. Furthermore, this study analyses the mechanism of SAM and IVMD in STFP, and proves the superiority of IVMD and SAM in signal decomposition with physical significance and weight allocation respectively. Therefore, the interpretability of the proposed neural network model is significantly improved. Eleven competitive neural networks forecasting methods are used as benchmarks and numerical studies based on actual traffic flow validate the accuracy and stability of the proposed IVMD-CNN-LSTM-SAM. Mean absolute percentage error and goodness of fit of the proposed model are 0.81% and 0.9968, respectively.

Wang, H. and Yue, P., 2020. Short-term traffic flow prediction of coastal cities based on entropy weight method. In: Yang, Y.; Mi, C.; Zhao, L., and Lam, S. (eds.), Global Topics and New Trends in Coastal Research: Port, Coastal and Ocean Engineering. Journal of Coastal Research, Special Issue No. 103, pp. 739–743. Coconut Creek (Florida), ISSN 0749-0208.The traditional short-term traffic flow prediction method has a high prediction error, so a short-term traffic flow prediction method based on entropy weight method is designed. First, identify the lost data, use the weighted average value of historical trend data and measured data to modify, and standardize the original data to collect the short-term traffic flow data of coastal cities. Then, statistical analysis of the original data set, using the minimum maximum standardization method, linear transformation of the original data, and extract the characteristic factors that affect the traffic flow. Finally, based on the correlation of traffic flow at urban intersections, the short-term traffic flow prediction of coastal cities is completed by using entropy weight method. In the experimental part, the prediction errors of the two methods in the period of 7:00-9:00 and 22:00-24:00 are compared. The comparison results show that the prediction errors of the short-term traffic flow prediction method based on entropy weight method in the period of 7:00-9:00 and 22:00-24:00 are small, which proves that the prediction method of this design has practical application significance.

Based on historical data, a new short-term traffic flow prediction model (ICEEMDAN-MPE-PSO-DELM-ARIMA) for intersections is proposed. In order to improve prediction accuracy, ICEEMDAN decomposition algorithm is applied on traffic flow time series to obtain multiple Intrinsic Mode Function (IMF) components. Then PSO-MPE algorithm is introduced to obtain the multi-scale permutation entropy values of each IMF component, to judge the randomness. According to the randomness, different prediction models are established. Prediction models based on PSO-DELM algorithm are established for IMF components with big randomness. ARIMA prediction models are established for IMF components with small randomness. In order to obtain the final predicted traffic flow values, multiple prediction results are added together. Finally, two actual intersections are selected to verify the proposed model. Results show that compared with other models, the proposed model has the smallest prediction errors and the best fitting effect with the real values, which can effectively improve prediction accuracy.

To address the issue of spatiotemporal illusion in short-term traffic flow prediction and deeply explore the underlying short-term traffic flow network characteristics, a traffic flow prediction model that combines long-term spatiotemporal heterogeneity with short-term spatiotemporal features is proposed. In the long-term spatiotemporal branch, the Transformer structure is employed, and a self-supervised masking mechanism is utilized to pretrain the heterogeneity in long-term temporal and spatial dimensions separately. Additionally, a spatiotemporal adaptive module is designed, which adapts to and guides short-term traffic flow prediction across time series and traffic flow networks. In the short-term spatiotemporal branch, a recurrent neural ordinary differential equation (ODE) module is devised. This module is capable of continuously and dynamically adjusting short-term spatiotemporal features, better capturing and exploring potential short-term spatiotemporal characteristics. Through multiple cycles, this module gradually and accurately extracts and compresses road network features, integrates the adapted spatiotemporal heterogeneity, and reconstructs future short-term traffic flows in the decoder. Experiments are conducted on four traffic flow and two traffic speed datasets, showing that compared to traditional time series models, the proposed model’s prediction accuracy indicators have relatively improved by 45.09%, 39.14%, and 0.47% on average; compared to recurrent neural network (RNN) series models, the improvements are 18.91%, 15.77%, and 0.18% on average; compared to graph convolution series models, the improvements are 21.31%, 16.65%, and 0.21% on average; and compared to Transformer series models, the improvements are 6.57%, 6.23%, and 0.05% on average. The model’s general applicability and good performance in transportation speed scenarios were verified through a multi-step experiment conducted on the transportation speed dataset.