In the real world, data missing is inevitable in traffic data collection due to detector failures or signal interference. However, missing traffic data imputation is non-trivial since traffic data usually contains bot...
详细信息
ISBN:
(纸本)9783030863838;9783030863821
In the real world, data missing is inevitable in traffic data collection due to detector failures or signal interference. However, missing traffic data imputation is non-trivial since traffic data usually contains both temporal and spatial characteristics with inherent complex relations. In each time interval, the traffic measurements collected in all spatial regions can be regarded as an image with more or fewer channels. Therefore, the traffic raster data over time can be learned as videos. In this paper, we propose a novel unsupervised generative neural network for traffic raster data imputation called STVAE, which works well robustly even under different missing rates. The core idea of our model is to discover more complex spatio-temporal representations inside the traffic data under the architecture of variational autoencoder (VAE) with Sylvester normalizing flows (SNFs). After transforming the traffic raster data into multi-channel videos, a Detection-and-Calibration Block (DCB), which extends 3D gated convolution and multi-attention mechanism, is proposed to sense, extract and calibrate more flexible and accurate spatio-temporal dependencies of the original data. The experiments are employed on three real-world traffic flow datasets and demonstrate that our network STVAE achieves the lowest imputation errors and outperforms state-of-the-art traffic data imputation models.
When dealing with high-dimensional data, such as in biometric, e-commerce, or industrial applications, it is extremely hard to capture the abnormalities in full space due to the curse of dimensionality. Furthermore, i...
详细信息
When dealing with high-dimensional data, such as in biometric, e-commerce, or industrial applications, it is extremely hard to capture the abnormalities in full space due to the curse of dimensionality. Furthermore, it is becoming increasingly complicated but essential to provide interpretations for outlier detection results in high-dimensional space as a consequence of the large number of features. To alleviate these issues, we propose a new model based on a variational autoencoder and Genetic Algorithm (VAEGA) for detecting outliers in subspaces of high-dimensional data. The proposed model employs a neural network to create a probabilistic dimensionality reduction variational autoencoder (VAE) that applies its low-dimensional hidden space to characterize the high-dimensional inputs. Then, the hidden vector is sampled randomly from the hidden space to reconstruct the data so that it closely matches the input data. The reconstruction error is then computed to determine an outlier score, and samples exceeding the threshold are tentatively identified as outliers. In the second step, a genetic algorithm (GA) is used as a basis for examining and analyzing the abnormal subspace of the outlier set obtained by the VAE layer. After encoding the outlier dataset's subspaces, the degree of anomaly for the detected subspaces is calculated using the redefined fitness function. Finally, the abnormal subspace is calculated for the detected point by selecting the subspace with the highest degree of anomaly. The clustering of abnormal subspaces helps filter outliers that are mislabeled (false positives), and the VAE layer adjusts the network weights based on the false positives. When compared to other methods using five public datasets, the VAEGA outlier detection model results are highly interpretable and outperform or have competitive performance compared to current contemporary methods.
The Mars Curiosity rover is frequently sending engineering and science data that goes through a pipeline of systems before reaching its final destination at the mission operations center making it prone to volume loss...
详细信息
The Mars Curiosity rover is frequently sending engineering and science data that goes through a pipeline of systems before reaching its final destination at the mission operations center making it prone to volume loss and data corruption. A ground data system analysis (GDSA) team is charged with the monitoring of this flow of information and the detection of anomalies in that data to request a re-transmission when necessary. So far, this task relied primarily on the expertise of GDSA analysts who, especially with new missions launching, require the assistance of an autonomous and effective detection tool. variational autoencoders are powerful deep neural networks that can learn to isolate such anomalies yet, as any deep network, they require an architectural search and fine-tuning to yield exploitable performance. Furthermore, this process needs to be repeated periodically to adjust to the changing flow of transmissions. This work presents Delta-MADS, a hybrid derivative-free optimization method designed to quickly produce efficient variational autoencoders in order to assist the GDSA team in their mission.
The availability of sparse and high dimensional consumer shopping data poses a challenge for researchers for accurate and efficient analysis. While deep learning models can handle such data, most of the results from t...
详细信息
The availability of sparse and high dimensional consumer shopping data poses a challenge for researchers for accurate and efficient analysis. While deep learning models can handle such data, most of the results from these models are uninterpretable. This greatly limits their value in applications aimed at better understanding multicategory shopping behavior and assisting managerial decision-making. Thus, a new approach is needed to analyze high dimensional consumer shopping data in an efficient and interpretable manner, with the potential to greatly strengthen decision support systems for managers especially at large consumer packaged goods firms and giant retailers. We propose a Bayesian deep learning approach based on variational autoencoder (VAE) designed to efficiently capture often complex and multifaceted interrelationships across items in the shopping baskets, in the form of substitute and complementary cross effects (which are observable and controllable) and coincidences (which are not). The benefits of the proposed approach are empirically demonstrated by applying our model to a high dimensional supermarket shopping dataset covering very large numbers of products at the stock-keeping unit (SKU) level, households, and shopping trips. We discuss implications for managerial decision-making and identify promising research directions.
Background and Objective: Epilepsy is one of the most common neurological disorders, whose development is typically detected via early seizures. Electroencephalogram (EEG) is prevalently employed for seizure identific...
详细信息
Background and Objective: Epilepsy is one of the most common neurological disorders, whose development is typically detected via early seizures. Electroencephalogram (EEG) is prevalently employed for seizure identification due to its routine and low expense collection. The stochastic nature of EEG makes manual seizure inspections laborsome, motivating automated seizure identification. The relevant literature focuses mostly on supervised machine learning. Despite their success, supervised methods require expert labels indicating seizure segments, which are difficult to obtain on clinically-acquired EEG. Thus, we aim to devise an unsupervised method for seizure identification on EEG. Methods: We propose the first fully-unsupervised deep learning method for seizure identification on raw EEG, using a variational autoencoder (VAE). In doing so, we train the VAE on recordings without seizures. As training captures non-seizure activity, we identify seizures with respect to the reconstruction errors at inference time. Moreover, we extend the traditional VAE training loss to suppress EEG artifacts. Our method does not require ground-truth expert labels indicating seizure segments or manual feature extraction. Results: We implement our method using the PyTorch library and execute experiments on an NVIDIA V100 GPU. We evaluate our method on three benchmark EEG datasets: (i) intracranial recordings from the University of Pennsylvania and the Mayo Clinic, (ii) scalp recordings from the Temple University Hospital of Philadelphia, and (iii) scalp recordings from the Massachusetts Institute of Technology and the Boston Children's Hospital. To assess performance, we report accuracy, precision, recall, Area under the Receiver Operating Characteristics Curve (AUC), and p-value under the Welch t-test for distinguishing seizure vs. non-seizure EEG windows. Our approach can successfully distinguish seizures from nonseizure activity, with up to 0.83 AUC on intracranial recordings. Moreove
Human activity recognition (HAR) became a challenging issue in recent years. In this paper, we propose a novel approach to tackle indistinguishable activity recognition based on human wearable sensors. Generally speak...
详细信息
Human activity recognition (HAR) became a challenging issue in recent years. In this paper, we propose a novel approach to tackle indistinguishable activity recognition based on human wearable sensors. Generally speaking, vision-based solutions struggle with low illumination environments and partial occlusion problems. In contrast, wearable inertial sensors can tackle this problem and avoid revealing personal privacy. We address the issue by building a multistage deep neural network framework that interprets accelerometer, gyroscope, and magnetometer data that provide useful information of human activities. Initially, the stage of variational autoencoders (VAE) can extract the crucial information from raw data of inertial measurement units (IMUs). Furthermore, the stage of generative adversarial networks (GANs) can generate more realistic human activities. Finally, the transfer learning method is applied to enhance the performance of the target domain, which builds a robust and effective model to recognize human activities.
The variational autoencoder (VAE) is a popular generative latent variable model that is often used for representation learning. Standard VAEs assume continuous-valued latent variables and are trained by maximization o...
详细信息
ISBN:
(纸本)9789082797039
The variational autoencoder (VAE) is a popular generative latent variable model that is often used for representation learning. Standard VAEs assume continuous-valued latent variables and are trained by maximization of the evidence lower bound (ELBO). Conventional methods obtain a differentiable estimate of the ELBO with reparametrized sampling and optimize it with Stochastic Gradient Descend (SGD). However, this is not possible if we want to train VAEs with discrete-valued latent variables, since reparametrized sampling is not possible. In this paper, we propose an easy method to train VAEs with binary or categorically valued latent representations. Therefore, we use a differentiable estimator for the ELBO which is based on importance sampling. In experiments, we verify the approach and train two different VAEs architectures with Bernoulli and categorically distributed latent representations on two different benchmark datasets.
Extracting pathology information embedded within surface optical properties in Spatial Frequency Domain Imaging (SFDI) datasets is still a rather cumbersome nonlinear translation problem, mainly constrained by intrasa...
详细信息
ISBN:
(数字)9781510628427
ISBN:
(纸本)9781510628410;9781510628427
Extracting pathology information embedded within surface optical properties in Spatial Frequency Domain Imaging (SFDI) datasets is still a rather cumbersome nonlinear translation problem, mainly constrained by intrasample and interpatient variability, as well as dataset size. The beta-variational autoencoder (beta-VAE) is a rather novel dimensionality reduction technique where a tractable set of latent low-dimensional embeddings can be obtained from a given dataset. These embeddings can then be sampled to synthesize new data, providing further insight into pathology variability as well as differentiability in terms of optical properties. Its applications for data classification and breast margin delineation are also discussed.
Human-robot interaction (HRI) is progressively addressing multi-party scenarios, where a robot interacts with more than one human user at the same time. Conversely, research in this area is still at an early stage for...
详细信息
ISBN:
(纸本)9781450394321
Human-robot interaction (HRI) is progressively addressing multi-party scenarios, where a robot interacts with more than one human user at the same time. Conversely, research in this area is still at an early stage for human-robot collaboration (HRC). The intervention of a robot in human collaboration could be helpful to handle mutual disturbances of workers operating at the same time on the same target object. Therefore, this work outlines design methodologies of non-dyadic human-robot collaborations to address concurrent human-human tasks in manufacturing applications. After this, preliminary results regarding a robotic agent's high-level understanding of such scenarios realised through a variational autoencoder trained by means of transfer learning are shown.
Wafer map analysis provides critical information for quality control and yield improvement tasks in semiconductor manufacturing. In particular, wafer patterns of gross failing areas (GFA) are important clues to identi...
详细信息
Wafer map analysis provides critical information for quality control and yield improvement tasks in semiconductor manufacturing. In particular, wafer patterns of gross failing areas (GFA) are important clues to identify the causes of relevant failures during the manufacturing process. In this work, a semi-supervised classification framework is proposed for wafer map analysis, and its application to wafer bin maps with GFA patterns classification is demonstrated. The Ladder network and the semi-supervised variational autoencoder are adopted to classify wafer bin maps in comparison with a standard convolutional neural network (CNN) model on two real-world datasets. The results have illustrated that two semi-supervised models are consistently and substantially better than the CNN model across various training data percentages by effective utilization of the unlabeled data. Active learning and pseudo labeling are also utilized to accelerate the learning curve.
暂无评论