Augmentation Algorithm Research Articles

Novel Data Augmentation Employing Multivariate Gaussian Distribution for Neural Network-Based Blood Pressure Estimation

In this paper, we propose a novel data augmentation technique employing multivariate Gaussian distribution (DA-MGD) for neural network (NN)-based blood pressure (BP) estimation, which incorporates the relationship between the features in a multi-dimensional feature vector to describe the correlated real-valued random variables successfully. To verify the proposed algorithm against the conventional algorithm, we compare the results in terms of mean error (ME) with standard deviation and Pearson correlation using 110 subjects contributed to the database (DB) which includes the systolic BP (SBP), diastolic BP (DBP), photoplethysmography (PPG) signal, and electrocardiography (ECG) signal. For each subject, 3 times (or 6 times) measurements are accomplished in which the PPG and ECG signals are recorded for 20 s. And, to compare with the performance of the BP estimation (BPE) using the data augmentation algorithms, we train the BPE model using the two-stage system, called the stacked NN. Since the proposed algorithm can express properly the correlation between the features than the conventional algorithm, the errors turn out lower compared to the conventional algorithm, which shows the superiority of our approach.

Bidirectional Mapping Augmentation Algorithm for Synthetic Images Based on Generative Adversarial Network

Bidirectional Mapping Augmentation Algorithm for Synthetic Images Based on Generative Adversarial Network

UAV Recognition Based on Micro-Doppler Dynamic Attribute-Guided Augmentation Algorithm

A micro-Doppler signature (m-DS) based on the rotation of drone blades is an effective way to detect and identify small drones. Deep-learning-based recognition algorithms can achieve higher recognition performance, but they needs a large amount of sample data to train models. In addition to the hovering state, the signal samples of small unmanned aerial vehicles (UAVs) should also include flight dynamics, such as vertical, pitch, forward and backward, roll, lateral, and yaw. However, it is difficult to collect all dynamic UAV signal samples under actual flight conditions, and these dynamic flight characteristics will lead to the deviation of the original features, thus affecting the performance of the recognizer. In this paper, we propose a small UAV m-DS recognition algorithm based on dynamic feature enhancement. We extract the combined principal component analysis and discrete wavelet transform (PCA-DWT) time–frequency characteristics and texture features of the UAV’s micro-Doppler signal and use a dynamic attribute-guided augmentation (DAGA) algorithm to expand the feature domain for model training to achieve an adaptive, accurate, and efficient multiclass recognition model in complex environments. After the training model is stable, the average recognition accuracy rate can reach 98% during dynamic flight.

An Improved VGG19 Transfer Learning Strip Steel Surface Defect Recognition Deep Neural Network Based on Few Samples and Imbalanced Datasets

The surface defects’ region of strip steel is small, and has various defect types and, complex gray structures. There tend to be a large number of false defects and edge light interference, which lead traditional machine vision algorithms to be unable to detect defects for various types of strip steel. Image detection techniques based on deep learning require a large number of images to train a network. However, for a dataset with few samples with category imbalanced defects, common deep learning neural network training tasks cannot be carried out. Based on rapid image preprocessing algorithms (improved gray projection algorithm, ROI image augmentation algorithm) and transfer learning theory, this paper proposes a set of processes for complete strip steel defect detection. These methods achieved surface rapid screening, defect feature extraction, sample dataset’s category balance, data augmentation, defect detection, and classification. Through verification of the mixed dataset, composed of the NEU surface dataset and dataset in this paper, the recognition accuracy of the improved VGG19 network in this paper reached 97.8%. The improved VGG19 network performs slightly better than the baseline VGG19 in six types of defects, but the improved VGG19 performs significantly better in the surface seams defects. The convergence speed and accuracy of the improved VGG19 network were taken into account, and the detection rate was greatly improved with few samples and imbalanced datasets. This paper also has practical value in terms of extending its method of strip steel defect detection to other products.

음성 인식과 텍스트 확장 알고리즘을 이용한 인물 성향 분석에 관한 연구 - 영화 내 주연 배우의 성향 중심으로

음성 인식과 텍스트 확장 알고리즘을 이용한 인물 성향 분석에 관한 연구 - 영화 내 주연 배우의 성향 중심으로

Differential diagnosis of ameloblastoma and odontogenic keratocyst by machine learning of panoramic radiographs

PurposeThe differentiation of the ameloblastoma and odontogenic keratocyst directly affects the formulation of surgical plans, while the results of differential diagnosis by imaging alone are not satisfactory. This paper aimed to propose an algorithm based on convolutional neural networks (CNN) structure to significantly improve the classification accuracy of these two tumors.MethodsA total of 420 digital panoramic radiographs provided by 401 patients were acquired from the Shanghai Ninth People’s Hospital. Each of them was cropped to a patch as a region of interest by radiologists. Furthermore, inverse logarithm transformation and histogram equalization were employed to increase the contrast of the region of interest (ROI). To alleviate overfitting, random rotation and flip transform as data augmentation algorithms were adopted to the training dataset. We provided a CNN structure based on a transfer learning algorithm, which consists of two branches in parallel. The output of the network is a two-dimensional vector representing the predicted scores of ameloblastoma and odontogenic keratocyst, respectively.ResultsThe proposed network achieved an accuracy of 90.36% (AUC = 0.946), while sensitivity and specificity were 92.88% and 87.80%, respectively. Two other networks named VGG-19 and ResNet-50 and a network trained from scratch were also used in the experiment, which achieved accuracy of 80.72%, 78.31%, and 69.88%, respectively.ConclusionsWe proposed an algorithm that significantly improves the differential diagnosis accuracy of ameloblastoma and odontogenic keratocyst and has the utility to provide a reliable recommendation to the oral maxillofacial specialists before surgery.

Regularizing Deep Networks With Semantic Data Augmentation.

Data augmentation is widely known as a simple yet surprisingly effective technique for regularizing deep networks. Conventional data augmentation schemes, e.g., flipping, translation or rotation, are low-level, data-independent and class-agnostic operations, leading to limited diversity for augmented samples. To this end, we propose a novel semantic data augmentation algorithm to complement traditional approaches. The proposed method is inspired by the intriguing property that deep networks are effective in learning linearized features, i.e., certain directions in the deep feature space correspond to meaningful semantic transformations, e.g., changing the background or view angle of an object. Based on this observation, translating training samples along many such directions in the feature space can effectively augment the dataset for more diversity. To implement this idea, we first introduce a sampling based method to obtain semantically meaningful directions efficiently. Then, an upper bound of the expected cross-entropy (CE) loss on the augmented training set is derived by assuming the number of augmented samples goes to infinity, yielding a highly efficient algorithm. In fact, we show that the proposed implicit semantic data augmentation (ISDA) algorithm amounts to minimizing a novel robust CE loss, which adds minimal extra computational cost to a normal training procedure. In addition to supervised learning, ISDA can be applied to semi-supervised learning tasks under the consistency regularization framework, where ISDA amounts to minimizing the upper bound of the expected KL-divergence between the augmented features and the original features. Although being simple, ISDA consistently improves the generalization performance of popular deep models (e.g., ResNets and DenseNets) on a variety of datasets, i.e., CIFAR-10, CIFAR-100, SVHN, ImageNet, and Cityscapes. Code for reproducing our results is available at https://github.com/blackfeather-wang/ISDA-for-Deep-Networks.

Effects of Data Augmentation Method Borderline-SMOTE on Emotion Recognition of EEG Signals Based on Convolutional Neural Network

In recent years, with the continuous development of artificial intelligence and brain-computer interface technology, emotion recognition based on physiological signals, especially electroencephalogram signals, has become a popular research topic and attracted wide attention. However, the imbalance of the data sets themselves, affective features’ extraction from electroencephalogram signals, and the design of classifiers with excellent performance, pose a great challenge to the subject. Motivated by the outstanding performance of deep learning approaches in pattern recognition tasks, we propose a method based on convolutional neural network with data augmentation method Borderline-synthetic minority oversampling technique. First, we obtain 32-channel electroencephalogram signals from DEAP data set, which is the standard data set of emotion recognition. Then, after data pre-processing, we extract features in frequency domain and data augmentation based on the data augmentation algorithm above for getting more balanced data. Finally, we train a one dimensional convolutional neural network for three classification on two emotional dimensions valence and arousal. Meanwhile, the proposed method is compared with some traditional machine learning methods and some existing methods by other researchers, which is proved to be effective in emotion recognition, and the average accuracy rate of 32 subjects on valence and arousal are 97.47% and 97.76% respectively. Compared with other existing methods, the performance of the proposed method with data augmentation algorithm Borderline-SMOTE shows its advantage in affective emotional recognition than that without Borderline-SMOTE.

Subsampled turbulence removal network

We present a deep-learning approach to restore a sequence of turbulence-distorted video frames from turbulent deformations and space-time varying blurs. Instead of requiring a massive training sample size in deep networks, we purpose a training strategy that is based on a new data augmentation method to model turbulence from a relatively small dataset. Then we introduce a subsampled method to enhance the restoration performance of the presented GAN model. The contributions of the paper is threefold: first, we introduce a simple but effective data augmentation algorithm to model the turbulence in real life for training in the deep network; Second, we firstly purpose the Wasserstein GAN combined with $\ell_1$ cost for successful restoration of turbulence-corrupted video sequence; Third, we combine the subsampling algorithm to filter out strongly corrupted frames to generate a video sequence with better quality.

A Novel Approach to Generate a Virtual Population of Human Coronary Arteries for In Silico Clinical Trials of Stent Design.

Goal: To develop a cardiovascular virtual population using statistical modeling and computational biomechanics. Methods: A clinical data augmentation algorithm is implemented to efficiently generate virtual clinical data using a real clinical dataset. An atherosclerotic plaque growth model is employed to 3D reconstructed coronary arterial segments to generate virtual coronary arterial geometries (geometrical data). Last, the combination of the virtual clinical and geometrical data is achieved using a methodology that allows for the generation of a realistic virtual population which can be used in in silico clinical trials. Results: The results show good agreement between real and virtual clinical data presenting a mean gof 0.1 ± 0.08. 400 virtual coronary arteries were generated, while the final virtual population includes 10,000 patients. Conclusions: The virtual arterial geometries are efficiently matched to the generated clinical data, both increasing and complementing the variability of the virtual population.

Cloud and Cloud Shadow Segmentation for Remote Sensing Imagery Via Filtered Jaccard Loss Function and Parametric Augmentation

Cloud and cloud shadow segmentation are fundamental processes in optical remote sensing image analysis. Current methods for cloud/shadow identification in geospatial imagery are not as accurate as they should, especially in the presence of snow and haze. This paper presents a deep learning-based framework for the detection of cloud/shadow in Landsat 8 images. Our method benefits from a convolutional neural network, Cloud-Net+ (a modification of our previously proposed Cloud-Net \cite{myigarss}) that is trained with a novel loss function (Filtered Jaccard Loss). The proposed loss function is more sensitive to the absence of foreground objects in an image and penalizes/rewards the predicted mask more accurately than other common loss functions. In addition, a sunlight direction-aware data augmentation technique is developed for the task of cloud shadow detection to extend the generalization ability of the proposed model by expanding existing training sets. The combination of Cloud-Net+, Filtered Jaccard Loss function, and the proposed augmentation algorithm delivers superior results on four public cloud/shadow detection datasets. Our experiments on Pascal VOC dataset exemplifies the applicability and quality of our proposed network and loss function in other computer vision applications.

A small samples training framework for deep Learning-based automatic information extraction: Case study of construction accident news reports analysis

Knowledge management is crucial for construction safety management. Widely collected and well-organized safety-related documents are recognized to be significant in raising the workers' security awareness and then to prevent hazards and accidents. To improve document processing efficiency, automatic information extraction plays an important role. However, currently, automatic information extraction modeling requires large scale training datasets. It is a big challenge for the engineering industry, especially for the fields which heavily rely on the experts’ knowledge. Limited data sources, and high time and labor costs make it not practical to establish a large-scale dataset. This work proposed a natural language data augmentation-based small samples training framework for automatic information extraction modeling. With the designed cross combination-based text data augmentation algorithm, the deep neural network can be employed to build up automatic information extraction models without large-scale raw data and manual annotations. Characters semantic coding is employed to avoid word segmentation and make sure that the framework can be utilized in different writing language systems. The BiLSTM-CRF model is adopted as the detection core to conduct character classification. Through a case study of two independent accident news report datasets analysis, the proposed framework has been validated. A reliable and robust automatic information extraction model can be established, even though with small samples training.

MeshCut data augmentation for deep learning in computer vision.

To solve overfitting in machine learning, we propose a novel data augmentation method called MeshCut, which uses a mesh-like mask to segment the whole image to achieve more partial diversified information. In our experiments, this strategy outperformed the existing augmentation strategies and achieved state-of-the-art results in a variety of computer vision tasks. MeshCut is also an easy-to-implement strategy that can efficiently improve the performance of the existing convolutional neural network models by a good margin without careful hand-tuning. The performance of such a strategy can be further improved by incorporating it into other augmentation strategies, which can make MeshCut a promising baseline strategy for future data augmentation algorithms.

An Artificial Neural Network Approach and a Data Augmentation Algorithm to Systematize the Diagnosis of Deep-Vein Thrombosis by Using Wells’ Criteria

The use of a back-propagation artificial neural network (ANN) to systematize the reliability of a Deep Vein Thrombosis (DVT) diagnostic by using Wells’ criteria is introduced herein. In this paper, a new ANN model is proposed to improve the Accuracy when dealing with a highly unbalanced dataset. To create the training dataset, a new data augmentation algorithm based on statistical data known as the prevalence of DVT of real cases reported in literature and from the public hospital is proposed. The above is used to generate one dataset of 10,000 synthetic cases. Each synthetic case has nine risk factors according to Wells’ criteria and also the use of two additional factors, such as gender and age, is proposed. According to interviews with medical specialists, a training scheme was established. In addition, a new algorithm is presented to improve the Accuracy and Sensitivity/Recall. According to the proposed algorithm, two thresholds of decision were found, the first one is 0.484, which is to improve Accuracy. The other one is 0.138 to improve Sensitivity/Recall. The Accuracy achieved is 90.99%, which is greater than that obtained with other related machine learning methods. The proposed ANN model was validated performing the k-fold cross validation technique using a dataset with 10,000 synthetic cases. The test was performed by using 59 real cases obtained from a regional hospital, achieving an Accuracy of 98.30%.

Improved Mosaic: Algorithms for more Complex Images

Data augmentation plays a vital role in deep learning, and image augmentation, as an important part of target detection and image classification, significantly improves the performance of the algorithm. The Mosaic data augmentation algorithm in YOLOv4 randomly selects 4 pictures from the train set and puts the contents of the 4 pictures into a synthetic picture that is directly used for training. This data augmentation method can improve the model’s recognition ability in complex backgrounds. In this paper, we improve the Mosaic data augmentation algorithm. After analyzing the synthesized picture area, it is divided into irregular grids, and a certain number of training set pictures are filled randomly, which further improves the synthesis ability and achieves a synthesis picture that can accommodate 6 and 9 training set pictures. After basic image processing methods such as zooming, flipping, and color gamut transformation, the model’s recognition ability under complex backgrounds is improved, and the accuracy of identifying small targets is improved. During the batch normalization operation, the data of 6 or 9 pictures can be calculated at the same time, which makes the model hyperparameter mini-batch need not be set very large, which reduces the GPU memory requirements.

Hiding in the Crowd: Federated Data Augmentation for On-Device Learning

To cope with the lack of on-device machine learning samples, this article presents a distributed data augmentation algorithm, coined federated data augmentation (FAug). In FAug, devices share a tiny fraction of their local data, i.e., seed samples, and collectively train a synthetic sample generator that can augment the local datasets of devices. To further improve FAug, we introduce a multihop-based seed sample collection method and an oversampling technique that mixes up collected seed samples. Both approaches enjoy the benefit from the crowd of devices, by hiding data privacy from preceding hops and feeding diverse seed samples. In the image classification tasks, simulations demonstrate that the proposed FAug frameworks yield stronger privacy guarantees, lower communication latency, and higher on-device ML accuracy.

INS/CNS Deeply Integrated Navigation Method of Near Space Vehicles.

Celestial navigation is required to improve the long-term accuracy preservation capability of near space vehicles. However, it takes a long time for traditional celestial navigation methods to identify the star map, which limits the improvement of the dynamic response ability. Meanwhile, the aero-optical effects caused by the near space environment can lead to the colorization of measurement noise, which affects the accuracy of the integrated navigation filter. In this paper, an INS/CNS deeply integrated navigation method, which includes a deeply integrated model and a second-order state augmented H-infinity filter, is proposed to solve these problems. The INS/CNS deeply integrated navigation model optimizes the attitude based on the gray image error function, which can estimate the attitude without star identification. The second-order state augmented H-infinity filter uses the state augmentation algorithm to whiten the measurement noise caused by the aero-optical effect, which can effectively improve the estimation accuracy of the H-infinity filter in the near space environment. Simulation results show that the proposed INS/CNS deeply integrated navigation method can reduce the computational cost by 50%, while the attitude accuracy is kept within 10” (3 ). The attitude root mean square of the second-order state augmented H-infinity filter does not exceed 5”, even when the parameter error increases to 50%, in the near space environment. Therefore, the INS/CNS deeply integrated navigation method can effectively improve the rapid response ability of the navigation system and the filtering accuracy in the near space environment, providing a reference for the future design of near space vehicle navigation systems.

Likelihood-Based Inference for Partially Observed Epidemics on Dynamic Networks

We propose a generative model and an inference scheme for epidemic processes on dynamic, adaptive contact networks. Network evolution is formulated as a link-Markovian process, which is then coupled to an individual-level stochastic susceptible-infectious-recovered model, to describe the interplay between the dynamics of the disease spread and the contact network underlying the epidemic. A Markov chain Monte Carlo framework is developed for likelihood-based inference from partial epidemic observations, with a novel data augmentation algorithm specifically designed to deal with missing individual recovery times under the dynamic network setting. Through a series of simulation experiments, we demonstrate the validity and flexibility of the model as well as the efficacy and efficiency of the data augmentation inference scheme. The model is also applied to a recent real-world dataset on influenza-like-illness transmission with high-resolution social contact tracking records. Supplementary materials for this article are available online.

Parameter-expanded data augmentation for analyzing correlated binary data using multivariate probit models.

Data augmentation has been commonly utilized to analyze correlated binary data using multivariate probit models in Bayesian analysis. However, the identification issue in the multivariate probit models necessitates a rigorous Metropolis-Hastings algorithm for sampling a correlation matrix, which may cause slow convergence and inefficiency of Markov chains. It is well-known that the parameter-expanded data augmentation, by introducing a working/artificial parameter or parameter vector, makes an identifiable model be non-identifiable and improves the mixing and convergence of data augmentation components. Therefore, we motivate to develop efficient parameter-expanded data augmentations to analyze correlated binary data using multivariate probit models. We investigate both the identifiable and non-identifiable multivariate probit models and develop the corresponding parameter-expanded data augmentation algorithms. We point out that the approaches, based on one non-identifiable model, circumvent a Metropolis-Hastings algorithm for sampling a correlation matrix and improve the convergence and mixing of correlation parameters; the identifiable model may produce the estimated regression parameters with smaller standard errors than the non-identifiable model does. We illustrate our proposed approaches using simulation studies and through the application to a longitudinal dataset from the Six Cities study.

Bi-directional Long Short-Term Memory model to analyze psychological effects on gamers

With the increasing popularity of android gaming applications on smart phones, detection of emotional states of hard-core gamers become the interest of study among psychologists. Although there exist a few interesting research works on the impact of video games over the child and adult group, most of them are only able to throw light on psychological aspects associated with the said cognitive task. The real-time detection of emotional states of the player while playing video game is still an unexplored area of research. The present work feels the void by proposing a novel scheme of detecting the emotional changes of human subject from their electroencephalographic (EEG) signal acquired during their engagement in playing video games. The problem is formulated in the settings of pattern classification, which involves four main steps: Data collection, pre-processing and artifact removal, feature extraction and classification.The novelty of the work lies in extracting the emotional content with a high recognition rate from the acquired EEG response using a deep learning algorithm. The primary contribution of the paper lies in efficient usage of a novel phase-sensitive Common Spatial Pattern algorithm for feature extraction and design of an attention-based Bi-directional Long Short-Term Memory (Bi-LSTM) network for classifying the emotional states of a video-game player into five classes: happiness, sadness, surprise, anger and neutral. Moreover, the scarcity of labeled data in EEG-based brain–computer​ interfacing (BCI) tasks is a serious issue while understanding the performance capabilities of the data-driven deep-learning models. Therefore, the present work also makes an attempt to handle the scarcity in the dimension of the extracted feature using a novel feature augmentation algorithm before feeding the feature-vector to the proposed Bi-LSTM network. Experiments undertaken yield productive and conclusive results that validate the efficacy of the proposed framework with the accuracy rate of 88.71%.