Recovering an image from a noisy observation is a key problem in signalprocessing. Recently, it has been shown that data-driven approaches employing convolutional neural networks can outperform classical model-based ...
详细信息
ISBN:
(纸本)9781538662496
Recovering an image from a noisy observation is a key problem in signalprocessing. Recently, it has been shown that data-driven approaches employing convolutional neural networks can outperform classical model-based techniques, because they can capture more powerful and discriminative features. However, since these methods are based on convolutional operations, they are only capable of exploiting local similarities without taking into account non-local self-similarities. In this paper we propose a convolutional neural network that employs graph-convolutional layers in order to exploit both local and non-local similarities. The graph-convolutional layers dynamically construct neighborhoods in the feature space to detect latent correlations in the feature maps produced by the hidden layers. The experimental results show that the proposed architecture outperforms classical convolutional neural networks for the denoising task.
With the widespread use of wearable electrocardiographic (ECG) devices, there’s a growing need for efficient processing of large-scale real-time data to detect cardiovascular diseases. Deep learning, known for its ac...
详细信息
Traditional time-series clustering methods usually perform poorly on high-dimensional data. However, image clustering using deep learning methods can complete image annotation and searches in large image databases wel...
详细信息
Traditional time-series clustering methods usually perform poorly on high-dimensional data. However, image clustering using deep learning methods can complete image annotation and searches in large image databases well. Therefore, this study aimed to propose a deep clustering model named GW_DC to convert one-dimensional time-series into two-dimensional images and improve cluster performance for algorithm users. The proposed GW_DC consisted of three processing stages: the image conversion stage, image enhancement stage, and image clustering stage. In the image conversion stage, the time series were converted into four kinds of two-dimensional images by different algorithms, including grayscale images, recurrence plot images, Markov transition field images, and Gramian Angular Difference Field images;this last one was considered to be the best by comparison. In the image enhancement stage, the signal components of two-dimensional images were extracted and processed by wavelet transform to denoise and enhance texture features. Meanwhile, a deep clustering network, combining convolutional neural networks with K-Means, was designed for well-learning characteristics and clustering according to the aforementioned enhanced images. Finally, six UCR datasets were adopted to assess the performance of models. The results showed that the proposed GW_DC model provided better results.
In this article, a new framework is proposed to address multi-class Motor imagery Brain-Computer Interface (MIBCI) problems containing a small portion of labeled datasets. In this framework, the combination of Indepen...
详细信息
In this article, a new framework is proposed to address multi-class Motor imagery Brain-Computer Interface (MIBCI) problems containing a small portion of labeled datasets. In this framework, the combination of Independent Component Analysis (ICA), multi-class Common Spatial Pattern (CSP), and a functional Application Programming Interface (API) model assumes a pivotal role. In the feature extraction stage of the work, a concatenated altered signal affected by spatial weights is proposed for each trial in three frequency ranges. This distribution of features can both provide suitable feature maps for augmentation, preparing data for the deep learning analysis, and underscore distinguishable features of MI classes. In the classification stage, spatial and temporal features are dominated by using the effective combination of a one-dimensional Convolutional neural Network (CNN) and a two-staged Bidirectional Long Short-Term Memory (BLSTM) in three branches containing different distributions of frequency. Given that, the model simultaneously learns past-to-future and future-to-past patterns in two stages. The experimental result on datasets 2a BCI-Competition IV illustrates that the proposed method can be liable, practical and more competitive than the other popular methods pointed out in this paper. All in all, the proposed framework can alleviate the issue of small portions of labeled datasets in MI problems.
Despite a great success in learning representation for image data, it is challenging to learn the stochastic latent features from natural language based on variational inference. The difficulty in stochastic sequentia...
详细信息
ISBN:
(纸本)9781479981311
Despite a great success in learning representation for image data, it is challenging to learn the stochastic latent features from natural language based on variational inference. The difficulty in stochastic sequential learning is due to the posterior collapse caused by an autoregressive decoder which is prone to be too strong to learn sufficient latent information during optimization. To compensate this weakness in learning procedure, a sophisticated latent structure is required to assure good convergence so that random features are sufficiently captured for sequential decoding. This study presents a new variational recurrent autoencoder (VRAE) for sequence reconstruction. There are two complementary encoders consisting of a long short-term memory (LSTM) and a pyramid bidirectional LSTM which are merged to discover the global and local dependencies in a hierarchical latent variable model, respectively. Experiments on Penn Treebank and Yelp 2013 demonstrate that the proposed hierarchical VRAE is able to learn the complementary representation as well as tackle the posterior collapse in stochastic sequential learning. The performance of recurrent autoencoder is substantially improved in terms of perplexity.
Just Noticeable Difference (JND) has many applica-tions in multimedia signalprocessing, especially for visual data processing up to date. It's generally defined as the minimum visual content changes that the huma...
详细信息
Graph neural networks (GNN s) are a powerful class of model for representation learning on relational data and graph-structured signal, such as brain connectivity graphs derived from neuroimaging. To date, existing wo...
详细信息
ISBN:
(数字)9781728163383
ISBN:
(纸本)9781665411844
Graph neural networks (GNN s) are a powerful class of model for representation learning on relational data and graph-structured signal, such as brain connectivity graphs derived from neuroimaging. To date, existing work applying graph learning methods to brain connectivity is limited to a single neuroimaging modality such as structural or functional MRI. In practice, the brain is best represented by multiple networks arising from different imaging modalities. We develop a gen-eral framework for jointly pooling multimodal graphs which share the same set of underlying nodes whilst differing in edge connectivity. Building on this approach, we propose a multimodal GNN (MM-GNN) model that incorporates mul-tiple types of neuroimaging-based brain connectivity. When applied to the task of classifying brain images from patients with schizophrenia and healthy control subjects, we observe that incorporating multimodal pooling dramatically improves performance over non-pooled networks and that MM-GNN matches or improves performance over multiple single-modal and non-GNN baselines. Finally, we demonstrate how our approach uses multimodal data to learn a unified, interpretable measure of the salience of individual brain regions of interest. In this way, MM-GNN represents a new method for leveraging diverse brain connectivity data to enhance the detection of mental health disorders and to understand their biological underpinnings.
Blur image classification is a key step to image recovery in imageprocessing. In this article, an ensemble convolution neural network (CNN) is designed to identify and classify four types of blur images: defocus blur...
详细信息
Blur image classification is a key step to image recovery in imageprocessing. In this article, an ensemble convolution neural network (CNN) is designed to identify and classify four types of blur images: defocus blur, Gaussian blur, haze blur, and motion blur. To achieve this, a two-stage pipeline, comprised of deep compression and ensemble technique, is proposed to enhance model discriminability without incurring additional computing burden. Specifically, our method first prunes the well-known networks, Alexnet and GoogleNet, by an appropriate compression ratio. The pruned networks are denoted as Simplified-Fast-Alexnet (SFA) and Simplified-Fast-GoogleNet (SFGN). Next, we employ an ensemble policy to integrate the SFA with SFGN as SFA+SFGN by assigning their respective weights based on a voting mechanism. In addition, to provide a benchmark set of blur image samples for training and testing blur classification models, we create a new public blur image dataset (available online at http://***/info/1092/***) containing 80,000+ patch-level, naturally blurred photographs, constructed using the improved super-pixel segmentation method, and 200,000+ artificially blurred images. Numerical experiments demonstrate the superior performance of the proposed approach in comparison with the original Alexnet and GoogleNet, as well as other state-of-the-art methods. (C) 2018 Elsevier B.V. All rights reserved.
Background: Finding analyzable metaphase chromosome images is an essential step in karyotyping which is a common task for clinicians to diagnose cancers and genetic disorders precisely. This step is tedious and time-c...
详细信息
Background: Finding analyzable metaphase chromosome images is an essential step in karyotyping which is a common task for clinicians to diagnose cancers and genetic disorders precisely. This step is tedious and time-consuming. Hence developing automated fast and reliable methods to assist clinical technicians becomes indispensable. Previous approaches include methods with feature extraction followed by rule or quality based classifiers, component analysis, and neural networks. methods: A two-stage automated metaphase-finding scheme, consisting of an imageprocessing based metaphase detection stage, and a deep convolutional neural network based selection stage is proposed. The first stage detects metaphase images from 10x scan of specimen slides. The selection stage, on the other hand, selects the analyzable ones among them. Results: The proposed scheme has a 99.33% true positive rate and 0.34% of the false positive rate of metaphase finding. Conclusion: This study demonstrates an effective scheme for the automated finding of analyzable metaphase images with high True positive and low False positive rates. (C) 2019 Elsevier Ltd. All rights reserved.
Tensor factorization is a useful technique for capturing the high-order interactions in data analysis. One assumption of tensor decompositions is that a predefined rank should be known in advance. However, the tensor ...
详细信息
Tensor factorization is a useful technique for capturing the high-order interactions in data analysis. One assumption of tensor decompositions is that a predefined rank should be known in advance. However, the tensor rank prediction is an NP-hard problem. The CANDECOMP/PARAFAC (CP) decomposition is a typical one. In this paper, we propose two methods based on convolutional neural network (CNN) to estimate CP tensor rank from noisy measurements. One applies CNN to the CP rank estimation directly. The other one adds a pre-decomposition for feature acquisition, which inputs rank-one components to CNN. Experimental results on synthetic and real-world datasets show the proposed methods outperforms state-of-the-art methods in terms of rank estimation accuracy.
暂无评论