Dilated Convolutional Neural Network Research Articles

GluNet: A Deep Learning Framework for Accurate Glucose Forecasting.

For people with Type 1 diabetes (T1D), forecasting of blood glucose (BG) can be used to effectively avoid hyperglycemia, hypoglycemia and associated complications. The latest continuous glucose monitoring (CGM) technology allows people to observe glucose in real-time. However, an accurate glucose forecast remains a challenge. In this work, we introduce GluNet, a framework that leverages on a personalized deep neural network to predict the probabilistic distribution of short-term (30-60 minutes) future CGM measurements for subjects with T1D based on their historical data including glucose measurements, meal information, insulin doses, and other factors. It adopts the latest deep learning techniques consisting of four components: data pre-processing, label transform/recover, multi-layers of dilated convolution neural network (CNN), and post-processing. The method is evaluated in-silico for both adult and adolescent subjects. The results show significant improvements over existing methods in the literature through a comprehensive comparison in terms of root mean square error (RMSE) ([Formula: see text] mg/dL) with short time lag ([Formula: see text] minutes) for prediction horizons (PH) = 30 mins (minutes), and RMSE ([Formula: see text] mg/dL) with time lag ([Formula: see text] mins) for PH = 60 mins for virtual adult subjects. In addition, GluNet is also tested on two clinical data sets. Results show that it achieves an RMSE ([Formula: see text]mg/dL) with time lag ([Formula: see text] mins) for PH = 30 mins and an RMSE ([Formula: see text] mg/dL) with time lag ([Formula: see text]mins) for PH = 60 mins. These are the best reported results for glucose forecasting when compared with other methods including the neural network for predicting glucose (NNPG), the support vector regression (SVR), the latent variable with exogenous input (LVX), and the auto regression with exogenous input (ARX) algorithm.

Sample Selection Based Change Detection with Dilated Network Learning in Remote Sensing Images

The change detection system plays an active role in remote sensing images by detecting the changes in the image environment. Changes may be occur frequently due to the seasonal factors such as deforestation, natural disorders, etc. The objective of this paper is to develop an effective approach to find the changes that were experienced over a time period. The changes detected by difference image (DI) method is not efficient because it has the drawbacks in speckle noise suppression and classification, also the quality of DI affects the change map. This paper proposes a sample selection-multiple dilated convolutional neural network for the detection of changed and unchanged areas. In this, multiple dilation rates are incorporated, that overcome the existing gridding problems in network. The feature map determined at fully connected network extracts the global information by widening the receptive field that covers all the regions without any missing portions in the image. Hence, this type of architecture improves the learning in deep network by detecting the changes accurately. Finally, the trained network model achieves a feature map from the two images with the classified result of changed and unchanged pixels. The accuracy of the proposed change detection result provides improved results as compared with the existing algorithms. The performance metrics like false positive, false negative, overall error, the percentage of correct classification and Kappa are used for the estimation of the proposed image change detection task. Experiments are conducted on seven real datasets, and the result shows the better detection result than the other existing approaches.

A deep separable neural network for human tissue identification in three-dimensional optical coherence tomography images

This research proposes a dilated depthwise separable network for human tissue identification using three-dimensional (3D) optical coherence tomography (OCT) images. Automatic human tissue identification has made it possible for fast pathological tissue analyses, detecting tissue changes over time, and efficiently making a precise therapy treatment plan. 3D medical image classification is a challenging task because of the indistinct tissue characteristics and computational efficiency. To address the challenge, a deep dilated depthwise separable convolutional neural network is proposed in this research. A depthwise separable architecture is introduced to improve parameter utilization efficiency. Dilated convolutions are applied to systematically aggregate multiscale contextual information and provide a large receptive field with a small number of trainable weights, which provide computational benefit. 2D convolutions are applied in the proposed model to enhance the computational efficiency. The constructed model is tested by performing a multi-class human thyroid tissue classification on 3D OCT images. For comparison, experimental results are obtained for texture feature-based shallow learning models and typical deep learning classification models. The results show that the proposed DDSCN model outperforms those state-of-art models and can improve accuracy by 3.2% compared with the best texture-based model and 2.27% compared with the best CNN model. The proposed deep model provides significant potential for the applicability of the deep learning technique to analyze medical images of human tissue while advancing the next generation of OCT-based real-time surgery image guidance.

Sparse attention based separable dilated convolutional neural network for targeted sentiment analysis

Long short-term memory networks (LSTM) and classical convolutional neural networks (CNN) are two critical methods for the task of targeted sentiment analysis, but LSTM are difficult to parallelize and time-inefficient, and classical CNN can only capture local semantic features. To this end, this paper first proposes a sparse attention based separable dilated convolutional neural network (SA-SDCCN), which consists of multichannel embedding layer, separable dilated convolution module, sparse attention layer, and output layer. Specifically, our work is mainly concentrated on the first three parts. In multichannel embedding layer, semantic and sentiment embeddings are incorporated into an embedding tensor, which builds richer representations over the input sequence. In separable dilated convolution module, long-range contextual semantic information is explored and multi-scale contextual semantic dependencies are aggregated simultaneously through diverse dilation rates. Moreover, the separable structure further reduces the model parameters. In sparse attention layer, sentiment-oriented components are noticed according to the features of specific target entity. Finally, some experiments on three benchmark datasets demonstrate that SA-SDCCN achieves comparable or even better performance than state-of-the-art methods in terms of higher parallelism and lower computational cost.

A Deep Learning Method for Bearing Fault Diagnosis through Stacked Residual Dilated Convolutions

Real-time monitoring and fault diagnosis of bearings are of great significance to improve production safety, prevent major accidents, and reduce production costs. However, there are three primary concerns in the current research, namely real-time performance, effectiveness, and generalization performance. In this paper, a deep learning method based on stacked residual dilated convolutional neural network (SRDCNN) is proposed for real-time bearing fault diagnosis, which is subtly combined by the dilated convolution, the input gate structure of long short-term memory network (LSTM) and the residual network. In the SRDCNN model, the dilated convolution is used to exponentially increase the receptive field of convolution kernel and extract features from the sample with more points, alleviating the influence of randomness. The input gate structure of LSTM could effectively remove noise and control the entry of information contained in the input sample. Meanwhile, the residual network is introduced to overcome the problem of vanishing gradients caused by the deeper structure of the neural network, hence improving the overall classification accuracy. The experimental results indicate that compared with three excellent models, the proposed SRDCNN model has higher denoising ability and better workload adaptability.

Cucumber leaf disease identification with global pooling dilated convolutional neural network

It is a challenging research topic to identify plant disease based on diseased leaf image processing techniques due to the complexity of the diseased leaf images. Deep learning models are promising for identifying plant disease based on leaf images and AlexNet is one of these models. Aiming at the problems of too many parameters of the AlexNet model and single feature scale, a global pooling dilated convolutional neural network (GPDCNN) is proposed in this paper for plant disease identification by combining dilated convolution with global pooling. Compared with the classical convolutional neural network (CNN) and AlexNet models, GPDCNN has three improvements: (1) the convolution receptive field are increased without increasing the computational complexity and without losing the discriminant formation by replacing fully connected layers with a global pooling layer; (2) dilated convolutional layer is employed to recover the spatial resolution without increasing the number of training parameters; (3) GPDCNN also integrates the merits of dilated convolution and global pooling. Experimental results on the datasets of six common cucumber leaf diseases demonstrate that the proposed model can effectively recognize cucumber diseases.

Image super resolution by dilated dense progressive network

Image super-resolution (SR) is an interesting topic in computer vision. However, it remains challenging to achieve high-resolution image from the corresponding low-resolution version due to inherent variability, high dimensionality, and small ground targets images. In this paper, a new model based on dilated convolutional neural network is proposed to improve the image resolution. Recently, deep learning methods have led to significant improvements and completely outpace other models. However, these methods have not fully exploited all the features of the original low-resolution image, because of complex imaging conditions and the degradation process. To address this issue, we proposed an effective model based on dilated dense network operations to accelerate deep networks for image SR, which support the exponential growth of the receptive field parallel by increasing the filter size. In particular, residual network and skip connections are used for deep recovery. The experimental evaluations on several datasets prove the efficiency and stability of the proposed model. The proposed model not only achieves state-of-the-art performance but also has more efficient computation.

Chinese Clinical Named Entity Recognition Using Residual Dilated Convolutional Neural Network With Conditional Random Field

Clinical named entity recognition (CNER) is a fundamental and crucial task for clinical and translation research. In recent years, deep learning methods have achieved significant success in CNER tasks. However, these methods depend greatly on recurrent neural networks (RNNs), which maintain a vector of hidden activations that are propagated through time, thus causing too much time to train models. In this paper, we propose a residual dilated convolutional neural network with the conditional random field (RD-CNN-CRF) for the Chinese CNER, which makes the model asynchronous in computation and thus speeding up the training period dramatically. To be more specific, Chinese characters and dictionary features are first projected into dense vector representations, then they are fed into the residual dilated convolutional neural network to capture contextual features. Finally, a conditional random field is employed to capture dependencies between neighboring tags and obtain the optimal tag sequence for the entire sequence. Computational results on the CCKS-2017 Task 2 benchmark dataset show that our proposed RD-CNN-CRF method competes favorably with state-of-the-art RNN-based methods both in terms of computational performance and training time.

A Dilated CNN Model for Image Classification

The dilated convolution algorithm, which is widely used for image segmentation, is applied in the image classification field in this paper. In many traditional image classification algorithms, convolution neural network (CNN) plays an important role. However, the classical CNN has the problem of consuming too much computing resources. To solve this problem, first, this paper proposed a dilated CNN model which is built through replacing the convolution kernels of traditional CNN by the dilated convolution kernels, and then, the dilated CNN model is tested on the Mnist handwritten digital recognition data set. Second, to solve the detail loss problem in the dilated CNN model, the hybrid dilated CNN (HDC) is built by stacking dilated convolution kernels with different dilation rates, and then the HDC model is tested on the wide-band remote sensing image data set of earth’s terrain. The results show that under the same environment, compared with the traditional CNN model, the dilated CNN model reduces the training time by 12.99% and improves the training accuracy by 2.86% averagely, compared with the dilated CNN model, the HDC model reduces the training time by 2.02% and improves the training and testing accuracy by 14.15% and 15.35% averagely. Therefore, the dilated CNN and HDC model proposed in this paper can significantly improve the image classification performance.

Branch Point Selection in RNA Splicing Using Deep Learning

Alternative splicing (AS) is a regulated process that takes place during gene expression by which a single gene may code for multiple proteins. This mechanism is controlled by a complex called spliceosome by which certain exons of a gene may be included in or excluded out from the final mRNA produced from that gene. In AS, at least three remarkable signals exist in introns and they are 5’ splice site (5’ss), the donor ss where GU nucleotides are more frequently present, 3’ss, the acceptor ss where AG nucleotides are more frequently present, and branch site. Generally, branch point site is located at 20 to 50 nucleotides upstream from the 3’ss. In this paper, we identify the branch point location using a computational model based on deep learning. We propose a hybrid model based on a combination of dilated convolution neural network and recurrent neural network. Integrating additional inputs to the raw RNA sequence has been studied such as conservation, binding energy, and di-nucleotide. The proposed model has been evaluated on two publicly available datasets and outperformed the current state-of-the-art methods. More specifically, the proposed model achieved for the first dataset 97.29% and 67.08% of the area under curve (ROC-AUC) and the area under precision recall curve (prAUC), respectively, for the second dataset 96.86% and 69.62% of ROC-AUC and prAUC, respectively. In addition, pathogenic variants have been studied by the proposed model and agreed with the reported ones biologically. To study RNA branch point selection, an easy-to-use Web server has been established for free access at: https://home.jbnu.ac.kr/NSCL/rnabps.htm .

Intelligent fault detection using raw vibration signals via dilated convolutional neural networks

Fault detection and diagnosis is critical to improve the reliability and availability in induction motors (IMs). Machine learning and deep learning techniques have been widely used in induction motor fault detection and diagnosis. In this paper, we propose a new deep learning model based on a dilated convolutional neural network (D-CNN) for detecting bearing faults in IMs. The proposed model works directly on raw vibration signals without any hand-crafted feature extraction process. Our model can incorporate global context without losing important local information by stacking dilated convolutions with an increasing width. Numerical results show that the proposed D-CNN is not only capable of classifying normal signals perfectly but also can achieve higher accuracy than conventional techniques under noisy environments.

Coronary artery centerline extraction in cardiac CT angiography using a CNN-based orientation classifier.

Coronary artery centerline extraction in cardiac CT angiography (CCTA) images is a prerequisite for evaluation of stenoses and atherosclerotic plaque. In this work, we propose an algorithm that extracts coronary artery centerlines in CCTA using a convolutional neural network (CNN). In the proposed method, a 3D dilated CNN is trained to predict the most likely direction and radius of an artery at any given point in a CCTA image based on a local image patch. Starting from a single seed point placed manually or automatically anywhere in a coronary artery, a tracker follows the vessel centerline in two directions using the predictions of the CNN. Tracking is terminated when no direction can be identified with high certainty. The CNN is trained using manually annotated centerlines in training images. No image preprocessing is required, so that the process is guided solely by the local image values around the tracker's location. The CNN was trained using a training set consisting of 8 CCTA images with a total of 32 manually annotated centerlines provided in the MICCAI 2008 Coronary Artery Tracking Challenge (CAT08). Evaluation was performed within the CAT08 challenge using a test set consisting of 24 CCTA test images in which 96 centerlines were extracted. The extracted centerlines had an average overlap of 93.7% with manually annotated reference centerlines. Extracted centerline points were highly accurate, with an average distance of 0.21 mm to reference centerline points. Based on these results the method ranks third among 25 publicly evaluated methods in CAT08. In a second test set consisting of 50 CCTA scans acquired at our institution (UMCU), an expert placed 5448 markers in the coronary arteries, along with radius measurements. Each marker was used as a seed point to extract a single centerline, which was compared to the other markers placed by the expert. This showed strong correspondence between extracted centerlines and manually placed markers. In a third test set containing 36 CCTA scans from the MICCAI 2014 Challenge on Automatic Coronary Calcium Scoring (orCaScore), fully automatic seeding and centerline extraction was evaluated using a segment-wise analysis. This showed that the algorithm is able to fully-automatically extract on average 92% of clinically relevant coronary artery segments. Finally, the limits of agreement between reference and automatic artery radius measurements were found to be below the size of one voxel in both the CAT08 dataset and the UMCU dataset. Extraction of a centerline based on a single seed point required on average 0.4 ± 0.1s and fully automatic coronary tree extraction required around 20s. The proposed method is able to accurately and efficiently determine the direction and radius of coronary arteries based on information derived directly from the image data. The method can be trained with limited training data, and once trained allows fast automatic or interactive extraction of coronary artery trees from CCTA images.

A novel dilated convolutional neural network model for road scene segmentation

Road scene understanding is one of the important modules in the field of autonomous driving. It can provide more information about roads and play an important role in building high-precision maps and real-time planning. Among them, semantic segmentation can assign category information to each pixe

RETRACTED: A novel dilated convolutional neural network model for road scene segmentation [EAI Endorsed Scal Inf Syst (2022), Online First

We, the Publisher, have retracted the following article: Yachao Zhang, Yuxia Yuan (2022). A novel dilated convolutional neural network model for road scene segmentation. EAI Endorsed Scal Inf Syst. http://dx.doi.org/10.4108/eai.27-1-2022.173164 The authors submitted the article to the Special

Monocular depth estimation with hierarchical fusion of dilated CNNs and soft-weighted-sum inference

Monocular depth estimation is very challenging in complex compositions depicting multiple objects of diverse scales. Albeit the recent great progress thanks to the deep convolutional neural networks, the state-of-the-art monocular depth estimation methods still fall short to handle such real-world challenging scenarios. In this paper, we propose a deep end-to-end learning framework to tackle these challenges, which learns the direct mapping from a color image to the corresponding depth map. First, we represent monocular depth estimation as a multi-category dense labeling task by contrast to the regression-based formulation. In this way, we could build upon the recent progress in dense labeling such as semantic segmentation. Second, we fuse different side-outputs from our front-end dilated convolutional neural network in a hierarchical way to exploit the multi-scale depth cues for monocular depth estimation, which is critical in achieving scale-aware depth estimation. Third, we propose to utilize soft-weighted-sum inference instead of the hard-max inference, transforming the discretized depth scores to continuous depth values. Thus, we reduce the influence of quantization error and improve the robustness of our method. Extensive experiments have been conducted on the Make3D, NYU v2, and KITTI datasets and superior performance have been achieved on NYU v2 and KITTI datasets compared with current state-of-the-art methods, which shows the superiority of our method. Furthermore, experiments on the NYU v2 dataset reveal that our classification based model is able to learn the probability distribution of depth.

MR-Only Brain Radiation Therapy: Dosimetric Evaluation of Synthetic CTs Generated by a Dilated Convolutional Neural Network

This work aims to facilitate a fast magnetic resonance (MR)-only workflow for radiation therapy of intracranial tumors. Here, we evaluate whether synthetic computed tomography (sCT) images generated with a dilated convolutional neural network (CNN) enable accurate MR-based dose calculations in the brain. We conducted a retrospective study of 52 patients with brain tumors who underwent both computed tomography (CT) and MR imaging for radiation therapy treatment planning. To generate the sCTs, a T1-weighted gradient echo MR sequence was selected from the clinical protocol for multiple types of brain tumors. sCTs were created for all 52 patients with a dilated CNN using 2-fold cross validation; in each fold, 26 patients were used for training and the remaining 26patients were used for evaluation. For each patient, the clinical CT-based treatment plan was recalculated on sCT. We calculated dose differences and gamma pass rates between CT- and sCT-based plans inside body and planning target volume. Geometric fidelity of the sCT and differences in beam depth and equivalent path length were assessed between both treatment plans. sCT generation took 1 minute per patient. Over the patient population, themean absolute error of the sCT within the intersection of body contours was 67± 11 HU (±1 standard deviation [SD], range: 51-117 HU), and the mean error was 13 ± 9 HU (±1 SD, range: -2 to 38 HU). Dosimetric analysis showed mean deviations of 0.00% ± 0.02% (±1 SD, range: -0.05 to 0.03) for dose within the body contours and -0.13% ± 0.39% (±1 SD, range: -1.43 to 0.80) inside the planning target volume. Mean γ1mm/1% was 98.8% ± 2.2% for doses >50% of the prescribed dose. The presented dilated CNN generated sCTs from conventional MR images without adding scan time to the acquisition. Dosimetric evaluation suggests that dose calculations performed on the sCTs are accurate and can therefore be used for MR-only intracranial radiation therapy treatment planning.

Fully automatic and robust segmentation of the clinical target volume for radiotherapy of breast cancer using big data and deep learning

PurposeTo train and evaluate a very deep dilated residual network (DD-ResNet) for fast and consistent auto-segmentation of the clinical target volume (CTV) for breast cancer (BC) radiotherapy with big data. MethodsDD-ResNet was an end-to-end model enabling fast training and testing. We used big data comprising 800 patients who underwent breast-conserving therapy for evaluation. The CTV were validated by experienced radiation oncologists. We performed a fivefold cross-validation to test the performance of the model. The segmentation accuracy was quantified by the Dice similarity coefficient (DSC) and the Hausdorff distance (HD). The performance of the proposed model was evaluated against two different deep learning models: deep dilated convolutional neural network (DDCNN) and deep deconvolutional neural network (DDNN). ResultsMean DSC values of DD-ResNet (0.91 and 0.91) were higher than the other two networks (DDCNN: 0.85 and 0.85; DDNN: 0.88 and 0.87) for both right-sided and left-sided BC. It also has smaller mean HD values of 10.5 mm and 10.7 mm compared with DDCNN (15.1 mm and 15.6 mm) and DDNN (13.5 mm and 14.1 mm). Mean segmentation time was 4 s, 21 s and 15 s per patient with DDCNN, DDNN and DD-ResNet, respectively. The DD-ResNet was also superior with regard to results in the literature. ConclusionsThe proposed method could segment the CTV accurately with acceptable time consumption. It was invariant to the body size and shape of patients and could improve the consistency of target delineation and streamline radiotherapy workflows.

Waveform Modeling and Generation Using Hierarchical Recurrent Neural Networks for Speech Bandwidth Extension

This paper presents a waveform modeling and generation method using hierarchical recurrent neural networks (HRNN) for speech bandwidth extension (BWE). Different from conventional BWE methods which predict spectral parameters for reconstructing wideband speech waveforms, this BWE method models and predicts waveform samples directly without using vocoders. Inspired by SampleRNN which is an unconditional neural audio generator, the HRNN model represents the distribution of each wideband or high-frequency waveform sample conditioned on the input narrowband waveform samples using a neural network composed of long short-term memory (LSTM) layers and feed-forward (FF) layers. The LSTM layers form a hierarchical structure and each layer operates at a specific temporal resolution to efficiently capture long-span dependencies between temporal sequences. Furthermore, additional conditions, such as the bottleneck (BN) features derived from narrowband speech using a deep neural network (DNN)-based state classifier, are employed as auxiliary input to further improve the quality of generated wideband speech. The experimental results of comparing several waveform modeling methods show that the HRNN-based method can achieve better speech quality and run-time efficiency than the dilated convolutional neural network (DCNN)-based method and the plain sample-level recurrent neural network (SRNN)-based method. Our proposed method also outperforms the conventional vocoder-based BWE method using LSTM-RNNs in terms of the subjective quality of the reconstructed wideband speech.

3D skeleton based action recognition by video-domain translation-scale invariant mapping and multi-scale dilated CNN

In this paper, we present an image classification approach to action recognition with 3D skeleton videos. First, we propose a video domain translation-scale invariant image mapping, which transforms the 3D skeleton videos to color images, namely skeleton images. Second, a multi-scale dilated convolutional neural network (CNN) is designed for the classification of the skeleton images. Our multi-scale dilated CNN model could effectively improve the frequency adaptiveness and exploit the discriminative temporal-spatial cues for the skeleton images. Even though the skeleton images are very different from natural images, we show that the fine-tuning strategy still works well. Furthermore, we propose different kinds of data augmentation strategies to improve the generalization and robustness of our method. Experimental results on popular benchmark datasets such as NTU RGB + D, UTD-MHAD, MSRC-12 and G3D demonstrate the superiority of our approach, which outperforms the state-of-the-art methods by a large margin.

Deep Dilation on Multimodality Time Series for Human Activity Recognition

Providing accurate information on people’s activities and behaviors plays an important role in innumerable applications, such as medical, security, and entertainment. In recent years, deep learning has been applied in human activity recognition, and achieved a better performance. However, if the spatial dependency of inter-sensors is considered, it is possible to enhance the discriminative ability. In this paper, we present a novel deep learning framework for human activity recognition problems. First, on the basis of our previous work, we utilize dilated convolutional neural network to extract features of inter-sensors and intra-sensors. Since the extracted features are local and short-temporal, it is necessary to utilize RNN to model the long-temporal dependencies. However, duo to that LSTM and GRU often rely on the completely previous computations, it will result in slow inference and hard convergence. Hence, inspired by the idea of dilation operation, we present a novel recurrent model to learn the temporal dependencies at different time scales. Then, at the topmost layer, a fully-connected layer with softmax function is utilized to generate a class probability distribution, and the predicted activity is obtained. Eventually, we evaluate the proposed framework in two open human activity datasets, OPPORTUNITY and PAMAP2. Results demonstrate that the proposed framework achieves a higher classification performance than the state-of-the-art methods. Moreover, it takes the least time to recognize an activity. Besides, it also performs faster and easier to converge in the training stage.