Improving the Traffic Data Imputation Accuracy Using Temporal and Spatial Information

The quality of traffic data is crucial for modern transportation planning and operations. However, data could be missing for various reasons. Hence, the data imputation approaches which aim at predicting/replacing the missing data or bad data have been considered very important. The traditional traffic data imputation approaches mainly focus on using different probability models or regression methods to impute data, and they only take limited temporal or spatial information as inputs. Thus, they are not very accurate especially for data with a high missing ratio. To overcome the weaknesses of previous approaches, this study proposes an innovative traffic data imputation method, which first transforms the raw data into spatial‐temporal images and then implements a deep‐learning method on the images. The key idea of this approach is developing a convolutional neural network (CNN)‐based context encoder to reconstruct the complete image from the missing source. To the best of the authors’ knowledge, this is the first time a CNN method has been incorporated for traffic data imputation. Experiments are conducted on three months of data from 256 loop detectors. Through comparison with two state‐of‐the‐art approaches, the results indicate that this new approach increases the imputation accuracy greatly and has a stable error distribution.

In recent studies, the Probabilistic Principal Component Analysis (PPCA) for imputing missing data was shown to be a good tool for traffic flow data processing. The PPCA method has two major benefits: it results in significantly smaller reconstruction errors and much less computation time, which make it outperform the conventional historical and regression imputing methods. In this paper, the possibility of applying more complex PPCA methods, e.g. Kernel Probabilistic Principal Component Analysis (KPPCA) or Mixed Probabilistic Principal Component Analysis (MPPCA), to impute missing data is explored. Considering reconstruction errors and computation time cost, test results show that the basic PPCA method is still our first choice in missing data imputing for traffic flow for online systems.

There is context-related temporal and spatial information in video. How to make better use of these information is the key to improving the accuracy and speed of video object detection. The method of processing spatial and temporal information through 3D convolution is better than that of processing spatial information first and then processing time dimension information. There are two reasons for this assertion. One is that the 3D convolution can process spatial information and temporal information at the same time which will not divide them into two parts in feature extraction. The other is that the human visual derivation mechanism needs to memorize multi frames and process a video as a whole. After the training test on NFS dataset, it is verified that 3D convolution has stronger ability to extract temporal and spatial information. In addition, this paper also proposes a downward sampling method by connecting the low-resolution layer and the high-resolution layer, which can effectively improve the FPN model and enrich its high-level feature representation. After training and testing with different number of downward sample modules on MOT16 dataset, the improvement of model performance by down-sampling module is verified. Finally, this paper combines the downward sample module with the 3D network, and obtains the best test results on the NFS dataset.

In standard Principal Component Analysis (PCA) comes one problem in addressing the set of incomplete data. The standard PCA procedure on incomplete data is to eliminate (listwise deletion procedure) or using the mean of the variable, this procedure may result in loss information from these observations. Another method used is to integrate Expectation Maximization (EM) to the method of Probabilistic Principal Component Analysis (PPCA). But PPCA can produce overfitting response prediction. In this study discussed the Variational Bayesian Principal Component Analysis (VBPCA) which is a method of development of PPCA method by incorporating prior information from the distribution of the principal components of the model parameters. From the simulation studies by eliminating the data through the concept of missing at random (MAR), obtained results that the value of the correlation scores principal components complete data with the principal component score predicted results PPCA method is superior when compared with VBPCA, as well as to the value of the correlation scores for the various percentages are generally incomplete data. However, judging from the size of a match between the response to predictions by the size normalized root mean square error of prediction (NRMSEP) VBPCA method produces better than PPCA.

The loss of traffic state data is a common problem in intelligent transportation system. To improve the imputation accuracy and robustness of road traffic data, a novel generative adversarial network for the imputation of road network traffic data is proposed in this paper, i.e., GAE-GAN-LSTM. The spatiotemporal characteristics were extracted using a improved graph auto-encoder (GAE), followed by a generative adversarial network (GAN) to generate the complete spatiotemporal characteristics on the basis of the missing features. The internal structure of the generator was a long short-term memory network (LSTM), and the internal structure of the discriminator was a fully connected neural network (FCN). Finally, the traffic state data could be recovered by the decoder of GAE. The experimental results revealed that the performance of the proposed method was better than that of the other methods at any data loss ratio considered. The main innovations of the proposed method include two aspects. One, an improved GAE for the imputation of road network data was presented by redefining the loss function of GAE, which could effectively extract the potential spatiotemporal features of a road network. Two, the GAN was used to generate the spatiotemporal characteristics of the traffic state data by using the strong data generation ability of GAN.KeywordsData imputationGraph auto-encoderGenerative adversarial networkRoad traffic state data

In the context of automotive driving assistance, this paper describes a generic (i.e. applicable to both vehicle interior and exterior scenes) vision based approach for scene content analysis. It makes use of temporal and spatial information from a stereoscopic sequence of images to localize objects and estimate their position and motion. The proposed method is divided into three steps. First, image features are selected, tracked and reconstructed in the 3D world space. Second, a clustering step is processed in the 5D space made of the positions and 2D motions parameters. The last step is devoted to clusters interpretation: it is out of the scope of the paper, however orientations are given to illustrate the capabilities of the proposed approach. The paper is organized as follows: first, the use of temporal and spatial information from a stereoscopic sequence is investigated. A state of the art of existing methods is presented. Then, a generic approach for object segmentation is proposed. Lastly, experimental results are presented.

Principal Component Analysis (PCA), also known as Proper Orthogonal Decomposition (POD), is one of the prevailing Reduced Order Modeling (ROM) techniques for aerodynamic data analysis. Along with the PCA, its variant for missing data, gappy POD, has recently found its application by transforming problems into missing data problems. In this paper, an alternative based on probability theory, i.e., Probabilistic Principal Component Analysis (PPCA), is employed for aerodynamic data reconstruction problems in lieu of the PCA and the gappy POD. Unlike the PCA, the PPCA generates spanning vectors of low-dimensional space iteratively with an Expectation Maximization (EM) algorithm. The combined algorithm, termed as Expectation-Maximization Probabilistic Principal Component Analysis (EM-PCA), can deal with not only complete data but also incomplete data sets, since the EM algorithm naturally incorporates missing data in its formulation. Therefore, the EM-PCA has potential to be a substitute for both PCA and gappy POD. For illustration, the EM-PCA is applied to complete as well as incomplete snapshot ensembles compiled from aerodynamic simulations, and acquired modes and restored data are verified with those obtained by the PCA and the gappy POD.

In order to evaluate the perceptual quality of images, a full-reference quality index, which is called principal component deviation (PCD), is presented. This research is motivated by the discovery that the filters learned by the probabilistic principal component analysis are closely resemble the neurons of the human visual system. These filters are used as a model of the visual system, which are trained on 100,000 color image patches of size 4×4 by probabilistic principal component analysis. The PCD relates the image quality with the deviation between two sets of features that extracted by the learned filters. Experimental results show that PCD has relatively low computational complexity and high correlation with subjective quality evaluations.

Gaussian mixture model (GMM) techniques are popular for speaker identification. Theoretically, each Gaussian function should have a full covariance matrix. However, the diagonal covariance matrix is usually used because the inverse of diagonal covariance matrix can be easily calculated via expectation maximization (EM) algorithm. This paper proposes a new probabilistic principal component analysis (PPCA) model for speaker identification. The full covariance of speaker’s data is considered. This model is originated from factor analysis theory. The probability distributions using PPCA are well defined. In particular, GMM and PPCA are found to be equivalent when using diagonal covariance matrix. In this study, we derive a novel PPCA model selection and establish models for different speakers. Applying PPCA model selection, we can dynamically determine the numbers of speech features and mixture components. Experiments show that PPCA achieves desirable speaker recognition performance with proper model regularization.

Modern industrial plants generally consist of multiple manufacturing units, and the local correlation within each unit can be used to effectively alleviate the effect of spurious correlation and meticulously reflect the operation status of the process system. Therefore, the local correlation, which is called spatial information here, should also be taken into consideration when developing the monitoring model. In this study, a cascaded monitoring network (MoniNet) method is proposed to develop the monitoring model with concurrent analytics of temporal and spatial information. By implementing convolutional operation to each variable, the temporal information that reveals dynamic correlation of process data and spatial information that reflects local characteristics within individual operation unit can be extracted simultaneously. For each convolutional feature, a submodel is developed and then all the submodels are integrated to generate a final monitoring model. Based on the developed model, the operation status of the newly collected sample can be identified by comparing the calculated statistics with their corresponding control limits. Similar to the convolutional neural network (CNN), the MoniNet can also expand its receptive field and capture deeper information by adding more convolutional layers. Besides, the filter selection and submodel development in MoniNet can be replaced to generalize the proposed network to many existing monitoring strategies. The performance of the proposed method is validated using two real industrial processes. The illustration results show that the proposed method can effectively detect process anomalies by concurrent analytics of temporal and spatial information.

Most of the existing hyperspectral image change detection methods based on deep learning utilize convolutional neural networks (CNN), recurrent neural networks (RNN), and long-short-term memory (LSTM). However, CNNs primarily focus on extracting spatial information from hyperspectral images, thereby failing to fully utilize channel and temporal information. In this paper, we propose a network called STSTNet that can jointly extract spatial, spectral, and temporal feature information. To achieve a richer feature representation space, we employ a spatial attention module to extract spatial feature information from feature maps, and a spectral transformer encoder structure to extract abundant spectral sequence information. Specifically, we design two temporal information extraction structures after each spatial information extraction structure or spectral information extraction structure to capture correlations and interactions within the hyperspectral images. Extensive experiments have been conducted on three hyperspectral datasets, and the experimental results demonstrate the network’s excellent performance.

Episodic memory involves the processing of spatial and temporal aspects of personal experiences. The lateral entorhinal cortex (LEC) plays an essential role in subserving memory. However, the mechanisms by which LEC integrates spatial and temporal information remain elusive. Here, we recorded LEC neurons while male rats performed one-dimensional tasks. Many LEC cells displayed spatial firing fields and demonstrated selectivity for traveling directions. Furthermore, some LEC neurons changed the firing rates of their spatial rate maps during a session (rate remapping). Importantly, this temporal modulation was consistent across sessions, even when the spatial environment was altered. Notably, the strength of temporal modulation was greater in LEC compared to other brain regions, such as the medial entorhinal cortex, CA1, and CA3. Thus, we demonstrate spatial rate mapping in LEC neurons, which may serve as a coding mechanism for temporal context, and allow for flexible multiplexing of spatial and temporal information.

The Acquisition and Long-Term Retention of Temporal, Spatial, and Item Information

Home Network-Attached Storage (NAS) provides an easy way for data sharing and backup among multiple consumer electronic devices in home networks. Because of large capacity and cost effectiveness, disks are widely adopted in home NAS devices. In addition, data writes are common for many home NAS devices since these devices are usually used for data storage and backup. A writeback policy selects which dirty buffers are to be flushed to the disk, which is critical to the system performance under write-intensive workloads. In this paper, an intelligent writeback policy for home NAS called TESA is proposed. In contrast to most existing writeback policies, TESA considers both temporal and spatial information of the dirty buffers to improve the writeback performance. Considering the temporal information helps to reduce the frequency of writing back recently used dirty buffers, and hence leads to reduced write traffic. Considering the spatial information causes a reduction in the overall seek time and rotation delay of the dirty buffer writeback. The TESA writeback policy was implemented on a NAS evaluation board running Linux kernel. The performance results shows that TESA yields a significant performance improvement (i.e., up to 33.1%) over the original writeback policy of Linux. Moreover, TESA outperforms the policy that considers only the temporal information of the dirty buffers by up to 21.0%, and it also has up to 10.9% performance improvement over a previous policy that considers both the temporal and spatial information.

Location prediction has attracted a significant amount of research effort. Given an object's recent movements and a future time, the goal of location prediction is to predict the location of this object at the future time specified. Prior works have elaborated on mining association relationships among regions, in which objects frequently appear, to predict locations. Association relationships among regions are represented as association rules. By exploring association relationships among regions, prior works are able to have a good accuracy for location prediction. However, with a large amount of trajectory data produced, a huge amount of association rules is expected. Furthermore, trajectory data has both the spatial and temporal information. To further enhance the accuracy of location prediction, one could utilize not only spatial information but also temporal information to estimate locations of objects. In this paper, we propose a spatial-temporal trajectory model (abbreviated as STT) to capture movement behaviors of objects. STT is represented as a probabilistic suffix tree with both spatial and temporal information of movements. Note that STT is able to discover sequential traversal relationships among regions and, for each region, STT derives the corresponding probabilities about the time when objects appear. With the nature of probabilistic suffix tree, we could use a compact structure to capture movement behavior of objects compared to association rules proposed. In light of STT, we further propose an algorithm to traverse STT for location prediction. By exploring both the spatial and temporal information of STT, the accuracy of location prediction is improved. To evaluate our proposed STT and prediction algorithm, we conduct experiments on the synthetic dataset generated from real datasets. Experimental results shows that our proposed STT is able to capture both spatial and temporal patterns of movement behaviors and, by exploring STT, our proposed prediction algorithm outperforms existing works.