Patient-Specific Deep Architectural Model for ECG Classification.

Classification of Bisyllabic Lexical Stress Patterns Using Deep Neural Networks

Deep neural networks for texture classification—A theoretical analysis

We investigate the use of Deep Neural Networks for the classification of image datasets where texture features are important for generating class-conditional discriminative representations. To this end, we first derive the size of the feature space for some standard textural features extracted from the input dataset and then use the theory of Vapnik-Chervonenkis dimension to show that hand-crafted feature extraction creates low-dimensional representations which help in reducing the overall excess error rate. As a corollary to this analysis, we derive for the first time upper bounds on the VC dimension of Convolutional Neural Network as well as Dropout and Dropconnect networks and the relation between excess error rate of Dropout and Dropconnect networks. The concept of intrinsic dimension is used to validate the intuition that texture-based datasets are inherently higher dimensional as compared to handwritten digits or other object recognition datasets and hence more difficult to be shattered by neural networks. We then derive the mean distance from the centroid to the nearest and farthest sampling points in an n-dimensional manifold and show that the Relative Contrast of the sample data vanishes as dimensionality of the underlying vector space tends to infinity.

Stacked autoencoder (SAE) is hard to achieve satisfactory performance, when input data are complex and non-stationary. Besides, the identification performance of recurrent neural network (RNN) may decrease rapidly under noisy environment. In order to deal with these problems, a novel hybrid deep neural network (DNN) based on stacked denoising autoencoder (SDAE) and gated recurrent unit neural network (GRUNN) is presented. First, the structure of the presented hybrid DNN is given. The hybrid DNN contains a SDAE, a GRUNN, and a softmax classifier. Then, the training algorithm based on action discovery (AD) is proposed to train the presented hybrid DNN. The experimental studies indicate the presented hybrid DNN processes strong anti-noise ability and adaptability to time-varying signals.

As a fundamental task in modern data mining, time series classification is powering mission-critical tasks including stock price prediction and network traffic analysis. Due to the non-linear structure of deep neural networks (DNN), deep learning has established as a promising solution to time series classification. However, the excessive learning capacity of DNNs may make them prone to threats of backdoor attacks, where an attacker embeds hidden functionalities (i.e., backdoor) to DNNs and activates the backdoor by specially-designed inputs (i.e., triggers). Despite extensive studies concerning backdoor attacks on image and text domains, there is little known about the vulnerability of DNN based time series classifiers against backdoor attacks. Due to the unique characteristics of time series data, most existing backdoor attack techniques fail to threaten time series classifiers. In this paper, through analyzing the key factors which influence the effectiveness of a backdoor, we systematize a list of practical principles for designing triggers on time series data. In this light, we propose a novel framework called TimeTrojan, which aims to learn to form the trigger pattern through a constrained multi-objective optimization. To solve the hereafter challenging optimization issue, we further design an iterative learning algorithm. Remarkably, the proposed framework is agnostic to a wide range of DNN classifiers. Extensive empirical results on 6 representative DNN classifiers and 6 real-world datasets validate the effectiveness of the proposed attack framework. In most cases, TimeTrojan successfully injects backdoors with 100% attack success rate without affecting the model accuracy on clean samples, which implies the complete control of the behavior of the DNN classifiers by the adversary.

Sensitive information such as credit card information, username, password and social security number etc, can be stolen using a fake page that imitates trusted website is called phishing. The attacker designs a similar webpage either by copying or making small manipulation to the legitimate page so that the online user cannot distinguish the legitimate and fake websites. A Deep Neural Network (DNN) was introduced to detect the phishing Uniform Resource Locator (URL). Initially, a 30-dimension feature vector was constructed based on URL-based features, Hypertext Markup Language (HTML)-based features and domain-based features. These features were processed in DNN to detect the phishing URL. However, the irrelevant, redundant and noisy features in the dataset increase the complexity of DNN classifier. So the feature selection is required for efficient phishing attack detection. But feature selection is a time-consuming process since it is an independent process. So in this paper, a feature vector is generated by DNN itself using Stacked Denoise Auto Encoder (SDAE). Moreover, the noisy data such as missing features affect the efficiency of phishing detection so the SDAE is trained to reconstruct a clean input feature vector. The initial input feature vector is corrupted by setting some feature vectors as zero. Then the corrupted feature vector is then mapped with basic auto encoder, to a hidden representation from which the input feature vector is reconstructed. The reconstructed features are given as input to DNN which selects the most relevant features and predicts the phishing URL. Hence the sparse feature representation of SDAE increases the classification accuracy of DNN. The experiments are conducted in Ham, Phishing Corpus and Phishload datasets in terms of accuracy, precision, recall and F-measure to prove the effectiveness of DNN-SDAE.

Due to the advent of modern embedded systems and mobile devices with constrained resources, there is a great demand for incredibly efficient deep neural networks for machine learning purposes. There is also a growing concern of privacy and confidentiality of user data within the general public when their data is processed and stored in an external server which has further fueled the need for developing such efficient neural networks for real-time inference on local embedded systems. The scope of our work presented in this paper is limited to image classification using a convolutional neural network. A Convolutional Neural Network (CNN) is a class of Deep Neural Network (DNN) widely used in the analysis of visual images captured by an image sensor, designed to extract information and convert it into meaningful representations for real-time inference of the input data. In this paper, we propose a neoteric variant of deep convolutional neural network architecture to ameliorate the performance of existing CNN architectures for real-time inference on embedded systems. We show that this architecture, dubbed CondenseNeXt, is remarkably efficient in comparison to the baseline neural network architecture, CondenseNet, by reducing trainable parameters and FLOPs required to train the network whilst maintaining a balance between the trained model size of less than 3.0 MB and accuracy trade-off resulting in an unprecedented computational efficiency.

Deep learning is a fast flattering highest development in the machine learning technique leading to augment performance in diverse medical applications. The tributary artificial intelligence is rapidly becoming metamorphic in the domain of Healthcare Technology rendering facility to analyze the medical data with high rapidity and exactness. Due to the overwhelming development of artificial intelligence (AI), the contemporary encroachments of deep learning employing advanced and futuristic deep learning methods has become a vivacious research domain for processing medical data in the health care industries. This paper presents a complete analysis about the deep structured learning based methods used in medical image analysis and also elucidate about the advancement of convolutional neural network which is a major class of deep neural networks used in medical image processing. It also gives insight about the fundamentals of deep learning methods and reviews in image interpretation. It also explains about the implementation of an integrated framework for medical image synthesis combining Speckle GAN and Di-Conv-AE-Net.

BackgroundIschemic stroke (IS) is one of the most common serious secondary diseases of atrial fibrillation (AF) within 1 year after its occurrence, both of which have manifestations of ischemia and hypoxia of the small vessels in the early phase of the condition. The fundus is a collection of capillaries, while the retina responds differently to light of different wavelengths. Predicting the risk of IS occurring secondary to AF, based on subtle differences in fundus images of different wavelengths, is yet to be explored. This study was conducted to predict the risk of IS occurring secondary to AF based on multi-spectrum fundus images using deep learning.MethodsA total of 150 AF participants without suffering from IS within 1 year after discharge and 100 IS participants with persistent arrhythmia symptoms or a history of AF diagnosis in the last year (defined as patients who would develop IS within 1 year after AF, based on fundus pathological manifestations generally prior to symptoms of the brain) were recruited. Fundus images at 548, 605, and 810 nm wavelengths were collected. Three classical deep neural network (DNN) models (Inception V3, ResNet50, SE50) were trained. Sociodemographic and selected routine clinical data were obtained.ResultsThe accuracy of all DNNs with the single-spectral or multi-spectral combination images at the three wavelengths as input reached above 78%. The IS detection performance of DNNs with 605 nm spectral images as input was relatively more stable than with the other wavelengths. The multi-spectral combination models acquired a higher area under the curve (AUC) scores than the single-spectral models.ConclusionsThe probability of IS secondary to AF could be predicted based on multi-spectrum fundus images using deep learning, and combinations of multi-spectrum images improved the performance of DNNs. Acquiring different spectral fundus images is advantageous for the early prevention of cardiovascular and cerebrovascular diseases. The method in this study is a beneficial preliminary and initiative exploration for diseases that are difficult to predict the onset time such as IS.

Image reconstruction for Electrical Impedance Tomography (EIT) is a highly nonlinear and ill-posed inverse problem. It requires the design and employment of feasible reconstruction methods capable to guarantee trustworthy image generation. Deep Neural Networks (DNN) have a powerful ability to express complex nonlinear functions. This research paper introduces a novel framework based on DNN aiming to achieve EIT image reconstruction. The proposed DNN model, comprises of the following two layers, namely: The Stacked Autoencoder (SAE) and the Logistic Regression (LR). It is trained using the large lab samples which are obtained by the COMSOL simulation software (a cross platform finite elements analysis solver). The relationship between the voltage measurement and the internal conductivity distribution is determined. The untrained voltage measurement samples are used as input to the trained DNN, and the output is an estimate for image reconstruction of the internal conductivity distribution. The results show that the proposed model can achieve reliable shape and size reconstruction. When white Gaussian noise with a signal-to-noise ratio of 30, 40 and 50 were added to test set, the proposed DNN structure still has good imaging results, which proved the anti-noise capability of the network. Furthermore, the network that was trained using simulation data sets, would be applied for the EIT image reconstruction based on the experimental data that were produced after preprocessing.

The main aim of this study is to test the ability of a deep neural network (DNN) to detect and to predict financial statement fraud (FSF) and non-financial statement fraud (NFSF). This study also intended to reduce increasing financial losses due to FSF and to improve the reliability of financial statements (FS) for various business and investment decision-making purposes. A supervised DNN application is introduced in this study, accompanied by an end-to-end automated process to improve the overall efficiency in detecting and predicting FSF and NFSF, using only annual reports of public listed companies from various industries with different market capital sizes. This study report focused on the application of DNN as the tech tool which leverages the traditional FSF-detecting techniques (financial ratios and financial risk indicators) as the basis to attempt to detect and predict the FSF. The findings of this study indicated that the proposed DNN model can predict and detect both FSF and NFSF with an accuracy score of 89.11%.

Recently, adversarial attacks have shown to lead the state-of-the-art deep neural networks (DNNs) to misclassification. However, most adversarial attacks are generated according to whether they are perceptual to human visual system, measured by geometric metrics such as the l2-norm, which ignores the common watermarks in cyber-physical systems. In this paper, we propose a fast adversarial watermark attack (FAWA) method based on fast differential evolution technique, which optimally superimposes a watermark on an image to fool DNNs. We also attempt to explain the reason why the attack is successful and propose two hypotheses on the vulnerability of DNN classifiers and the influence of the watermark attack on higher-layer features extraction respectively. In addition, we propose two countermeasure methods against FAWA based on random rotation and median filtering respectively. Experimental results show that our method achieves 41.3% success rate in fooling VGG-16 and have good transferability. Our approach is also shown to be effective in deceiving deep learning as a service (DLaaS) systems as well as the physical world. The proposed FAWA, hypotheses, and the countermeasure methods, provide a timely help for DNN designers to gain some knowledge of model vulnerability while designing DNN classifiers and related DLaaS applications.

Deep Neural Networks (DNN) have strong capabilities of memories, feature identifications and automatic analyses, solving various complex problems. However, DNN classifiers have obvious fragility that adding several unnoticeable perturbations to the original examples will lead to the errors in the classifier identification. In the field of communications, the adversarial examples will greatly reduce the accuracy of the signal identification, causing great information security risks. Considering the adversarial examples pose a serious threat to the security of the DNN models, studying their generation mechanisms and testing their attack effects are critical to ensuring the information security of the communication networks. This paper will study the generation of the adversarial examples and the influences of the adversarial examples on the accuracy of the DNN-based communication signal identification. Meanwhile, this paper will study the influences of the adversarial examples under the white-box models and black-box models, and explore the adversarial attack influences of the factors such as perturbation levels and iterative steps. The insights of this study would be helpful for ensuring the security of information networks and designing robust DNN communication networks.

Deep learning architectures have emerged as powerful function approximators in a broad spectrum of complex representation learning tasks, such as, computer vision, natural language processing and collaborative filtering. These architectures bear a high potential to learn the intrinsic structure of data and extract valuable insights. Despite the surge in the development of state-of-the-art intelligent systems using the deep neural networks (DNNs), these systems have found to be vulnerable to adversarial examples produced by adding a small-magnitude of perturbations. Such adversarial examples are adept at misleading the DNN classifiers. In the past, different attack strategies have been proposed to produce adversarial examples in the digital, physical, and transform domain, but the likelihood to generate perceptually realistic adversarial examples require more research efforts. In this paper, we present a novel approach to produce adversarial examples by combining the single-shot fast gradient sign method (FGSM) and spatial, as well as, transform domain image processing techniques. The resulted perturbations neutralize the impact of low-intensity based regions, thus, instilling the noise only in the selective high-intensity regions of the input image. While combining the customized perturbation with one-step FGSM perturbation in an un-targeted black-box attack scenario, the proposed approach successfully fools state-of-the-art DNN classifiers with 99% adversarial examples being misclassified on the ImageNet validation dataset.

Classification using deep learning neural networks for brain tumors