SRResNet Performance Enhancement Using Patch Inputs and Partial Convolution-Based Padding

Due to highly underdetermined nature of Single Image Super-Resolution (SISR) problem, deep learning neural networks are required to be more deeper to solve the problem effectively. One of deep neural networks successful in the Super-Resolution (SR) problem is ResNet which can render the capability of deeper networks with the help of skip connections. However, zero padding (ZP) scheme in the network restricts benefits of skip connections in SRResNet and its performance as the ratio of the number of pure input data to that of zero padded data increases. In this paper. we consider the ResNet with Partial Convolution based Padding (PCP) instead of ZP to solve SR problem. Since training of deep neural networks using patch images is advantageous in many aspects such as the number of training image data and network complexities, patch image based SR performance is compared with single full image based one. The experimental results show that patch based SRResNet SR results are better than single full image based ones and the performance of deep SRResNet with PCP is better than the one with ZP.     

Acoustic Emission Recognition Based on a Two-Streams Convolutional Neural Network

The Convolutional Neural Network (CNN) is a widely used deep neural network. Compared with the shallow neural network, the CNN network has better performance and faster computing in some image recognition tasks. It can effectively avoid the problem that network training falls into local extremes. At present, CNN has been applied in many different fields, including fault diagnosis, and it has improved the level and efficiency of fault diagnosis. In this paper, a two-streams convolutional neural network (TCNN) model is proposed. Based on the short-time Fourier transform (STFT) spectral and Mel Frequency Cepstrum Coefficient (MFCC) input characteristics of two-streams acoustic emission (AE) signals, an AE signal processing and classification system is constructed and compared with the traditional recognition methods of AE signals and traditional CNN networks. The experimental results illustrate the effectiveness of the proposed model. Compared with single-stream convolutional neural network and a simple Long Short-Term Memory (LSTM) network, the performance of TCNN which combines spatial and temporal features is greatly improved, and the accuracy rate can reach 100% on the current database, which is 12% higher than that of single-stream neural network.     

Research on the Application of Super Resolution Reconstruction Algorithm for Underwater Image

Underwater imaging is widely used in ocean, river and lake exploration, but it is affected by properties of water and the optics. In order to solve the lower-resolution underwater image formed by the influence of water and light, the image super-resolution reconstruction technique is applied to the underwater image processing. This paper addresses the problem of generating super-resolution underwater images by convolutional neural network framework technology. We research the degradation model of underwater images, and analyze the lower-resolution factors of underwater images in different situations, and compare different traditional super-resolution image reconstruction algorithms. We further show that the algorithm of super-resolution using deep convolution networks (SRCNN) which applied to super-resolution underwater images achieves good results.     

Identifying Materials of Photographic Images and Photorealistic Computer Generated Graphics Based on Deep CNNs

Currently, some photorealistic computer graphics are very similar to photographic images. Photorealistic computer generated graphics can be forged as photographic images, causing serious security problems. The aim of this work is to use a deep neural network to detect photographic images (PI) versus computer generated graphics (CG). In existing approaches, image feature classification is computationally intensive and fails to achieve real-time analysis. This paper presents an effective approach to automatically identify PI and CG based on deep convolutional neural networks (DCNNs). Compared with some existing methods, the proposed method achieves real-time forensic tasks by deepening the network structure. Experimental results show that this approach can effectively identify PI and CG with average detection accuracy of 98%.     

A Novel Combinational Convolutional Neural Network for Automatic Food-Ingredient Classification

With the development of deep learning and Convolutional Neural Networks (CNNs), the accuracy of automatic food recognition based on visual data have significantly improved. Some research studies have shown that the deeper the model is, the higher the accuracy is. However, very deep neural networks would be affected by the overfitting problem and also consume huge computing resources. In this paper, a new classification scheme is proposed for automatic food-ingredient recognition based on deep learning. We construct an up-to-date combinational convolutional neural network (CBNet) with a subnet merging technique. Firstly, two different neural networks are utilized for learning interested features. Then, a well-designed feature fusion component aggregates the features from subnetworks, further extracting richer and more precise features for image classification. In order to learn more complementary features, the corresponding fusion strategies are also proposed, including auxiliary classifiers and hyperparameters setting. Finally, CBNet based on the well-known VGGNet, ResNet and DenseNet is evaluated on a dataset including 41 major categories of food ingredients and 100 images for each category. Theoretical analysis and experimental results demonstrate that CBNet achieves promising accuracy for multi-class classification and improves the performance of convolutional neural networks.     

Embedding Image Through Generated Intermediate Medium Using Deep Convolutional Generative Adversarial Network

Deep neural network has proven to be very effective in computer vision fields. Deep convolutional network can learn the most suitable features of certain images without specific measure functions and outperform lots of traditional image processing methods. Generative adversarial network (GAN) is becoming one of the highlights among these deep neural networks. GAN is capable of generating realistic images which are imperceptible to the human vision system so that the generated images can be directly used as intermediate medium for many tasks. One promising application of using GAN generated images would be image concealing which requires the embedded image looks like not being tampered to human vision system and also undetectable to most analyzers. Texture synthesizing has drawn lots of attention in computer vision field and is used for image concealing in steganography and watermark. The traditional methods which use synthesized textures for information hiding mainly select features and mathematic functions by human metrics and usually have a low embedding rate. This paper takes advantage of the generative network and proposes an approach for synthesizing complex texture-like image of arbitrary size using a modified deep convolutional generative adversarial network (DCGAN), and then demonstrates the feasibility of embedding another image inside the generated texture while the difference between the two images is nearly invisible to the human eyes.     

Classification and Research of Skin Lesions Based on Machine Learning

Classification of skin lesions is a complex identification challenge. Due to the wide variety of skin lesions, doctors need to spend a lot of time and effort to judge the lesion image which zoomed through the dermatoscopy. The diagnosis which the algorithm of identifying pathological images assists doctors gets more and more attention. With the development of deep learning, the field of image recognition has made longterm progress. The effect of recognizing images through convolutional neural network models is better than traditional image recognition technology. In this work, we try to classify seven kinds of lesion images by various models and methods of deep learning, common models of convolutional neural network in the field of image classification include ResNet, DenseNet and SENet, etc. We use a fine-tuning model with a multi-layer perceptron, by training the skin lesion model, in the validation set and test set we use data expansion based on multiple cropping, and use five models’ ensemble as the final results. The experimental results show that the program has good results in improving the sensitivity of skin lesion diagnosis.     

Adversarial Attacks on License Plate Recognition Systems

The license plate recognition system (LPRS) has been widely adopted in daily life due to its efficiency and high accuracy. Deep neural networks are commonly used in the LPRS to improve the recognition accuracy. However, researchers have found that deep neural networks have their own security problems that may lead to unexpected results. Specifically, they can be easily attacked by the adversarial examples that are generated by adding small perturbations to the original images, resulting in incorrect license plate recognition. There are some classic methods to generate adversarial examples, but they cannot be adopted on LPRS directly. In this paper, we modify some classic methods to generate adversarial examples that could mislead the LPRS. We conduct extensive evaluations on the HyperLPR system and the results show that the system could be easily attacked by such adversarial examples. In addition, we show that the generated images could also attack the black-box systems; we show some examples that the Baidu LPR system also makes incorrect recognitions. We hope this paper could help improve the LPRS by realizing the existence of such adversarial attacks.     

Adversarial Learning for Distant Supervised Relation Extraction

Recently, many researchers have concentrated on using neural networks to learn features for Distant Supervised Relation Extraction (DSRE). These approaches generally use a softmax classifier with cross-entropy loss, which inevitably brings the noise of artificial class NA into classification process. To address the shortcoming, the classifier with ranking loss is employed to DSRE. Uniformly randomly selecting a relation or heuristically selecting the highest score among all incorrect relations are two common methods for generating a negative class in the ranking loss function. However, the majority of the generated negative class can be easily discriminated from positive class and will contribute little towards the training. Inspired by Generative Adversarial Networks (GANs), we use a neural network as the negative class generator to assist the training of our desired model, which acts as the discriminator in GANs. Through the alternating optimization of generator and discriminator, the generator is learning to produce more and more discriminable negative classes and the discriminator has to become better as well. This framework is independent of the concrete form of generator and discriminator. In this paper, we use a two layers fully-connected neural network as the generator and the Piecewise Convolutional Neural Networks (PCNNs) as the discriminator. Experiment results show that our proposed GAN-based method is effective and performs better than state-of-the-art methods.     

基于深度神经网络的货架商品识别方法

目的为快速统计货架商品信息,提出一种基于深度神经网络的货架商品自动识别方法。方法摄像头采集的货架商品图像经过深度神经网络算法处理,得到了图像中商品的SKU和位置。针对货架商品识别这种密集检测场景,文中方法改进了通用深度神经网络目标检测算法,将算法分为检测和分类2个阶段且重新设计了部分网络结构。最后,将文中方法和传统货架商品识别方法以及通用深度神经网络目标检测方法进行了比较。结果实验证明该方法的检测阶段的模型平均正确率达到96.5%,分类阶段的分类准确率达到99.9%。整图测试的查准率为97.56%,查全率为99.26%。结论相较于以往使用传统的目标检测模型进行货架商品识别以及使用SIFT等人工算子提取特征并分类识别商品具体SKU,文中方法的商品检出率和分类准确率都有了大幅度的提升,具有很好的应用潜力。     

Automatic Detection of Aortic Dissection Based on Morphology and Deep Learning

Aortic dissection (AD) is a kind of acute and rapidly progressing cardiovascular disease. In this work, we build a CTA image library with 88 CT cases, 43 cases of aortic dissection and 45 cases of health. An aortic dissection detection method based on CTA images is proposed. ROI is extracted based on binarization and morphology opening operation. The deep learning networks (InceptionV3, ResNet50, and DenseNet) are applied after the preprocessing of the datasets. Recall, F1-score, Matthews correlation coefficient (MCC) and other performance indexes are investigated. It is shown that the deep learning methods have much better performance than the traditional method. And among those deep learning methods, DenseNet121 can exceed other networks such as ResNet50 and InceptionV3.     

A Recommendation System Based on Fusing Boosting Model and DNN Model

In recent years, the models combining traditional machine learning with the deep learning are applied in many commodity recommendation practices. It has been proved better performance by the means of the neural network. Feature engineering has been the key to the success of many click rate estimation model. As we know, neural networks are able to extract high-order features automatically, and traditional linear models are able to extract low-order features. However, they are not necessarily efficient in learning all types of features. In traditional machine learning, gradient boosting decision tree is a typical representative of the tree model, which can construct new features related before and after tree. Convolutional neural networks have a better perception of local features. In this paper, we take advantage of convolutional networks to capture the local features. The features are constructed by the node leaf of gradient boosting decision tree. This paper employs the tree leaf node to mine the user behavior path features, and uses the deep model to extract the user abstract features. Based on a Kaggle competition, our model performs better in the test data than any other model.     

Landmark-based multi-region ensemble convolutional neural networks for bone age assessment

Bone age assessment based on hand X-ray imaging is important in pediatry medicine. At present, prediction of bone age is mainly performed by the manual comparison with the existing atlas. To develop an automatic regression framework based on deep learning with high performance and efficiency. A landmark-based multi-region convolutional neural networks for automatic bone age assessment based on left hand X-ray images was proposed. The deep alignment network localized multiple landmarks distributed over the hand, and cropped the local regions to establish the multi-region ensemble convolutional neural networks with different sub-network combinations. The modified loss function and the optimized bone sub-regions were applied to train the networks. The experiments on Digital Hand Atlas Database revealed that the mean absolute error of bone age assessment was 0.52 ± 0.25 years. It is the first study to predict bone age using deep learning methods throughout the entire process including image preprocessing, landmark localization and bone age predication. The proposed method outperformed most of the existing state-of-the-art deep learning methods and achieved good results compared with the expert's experience. It can improve the efficiency of the medical doctors while minimizing the subjective errors.     

Novel DDoS Feature Representation Model Combining Deep Belief Network and Canonical Correlation Analysis

Distributed denial of service (DDoS) attacks launch more and more frequently and are more destructive. Feature representation as an important part of DDoS defense technology directly affects the efficiency of defense. Most DDoS feature extraction methods cannot fully utilize the information of the original data, resulting in the extracted features losing useful features. In this paper, a DDoS feature representation method based on deep belief network (DBN) is proposed. We quantify the original data by the size of the network flows, the distribution of IP addresses and ports, and the diversity of packet sizes of different protocols and train the DBN in an unsupervised manner by these quantified values. Two feedforward neural networks (FFNN) are initialized by the trained deep belief network, and one of the feedforward neural networks continues to be trained in a supervised manner. The canonical correlation analysis (CCA) method is used to fuse the features extracted by two feedforward neural networks per layer. Experiments show that compared with other methods, the proposed method can extract better features.     

基于KAB模型的设计要素组合推导

目的 为了提高产品感性设计开发效率及意象匹配精度,采用定性和定量相结合的方法,提出一种基于层次分析法(AHP)与BP神经网络相结合的产品意象设计要素组合推导方法。方法 首先通过网络爬虫和亲和图法建立产品意象及造型数据库,以获得意象和设计要素;其次运用AHP构建产品层次结构模型及判断矩阵,计算意象及设计要素的权重系数;接着,基于形态拆解法与权重结果获得设计要素类型及优化组合编码,再运用语义差异法(SD)获取组合编码的用户感性意象均值;最后通过感性工学和AHP-BP神经网络构建KAB关键设计要素组合预测模型。结果 基于此模型预测四旋翼无人机设计方案,应用逼近理想解排序法(TOPSIS)对其进行验证评价,结果表明通过模型计算能够得到与目标感性意象高度匹配的设计要素组合编码。结论 基于此模型能够快速获得客观准确的产品意象造型设计要素组合,提高产品设计开发过程的效率。     

Deep convolutional neural networks for eigenvalue problems in mechanics

We show that deep convolutional neural networks (CNNs) can massively outperform traditional densely connected neural networks (NNs) (both deep or shallow) in predicting eigenvalue problems in mechanics. In this sense, we strike out in a new direction in mechanics computations with strongly predictive NNs whose success depends not only on architectures being deep but also being fundamentally different from the widely used to date. We consider a model problem: predicting the eigenvalues of one-dimensional (1D) and two-dimensional (2D) phononic crystals. For the 1D case, the optimal CNN architecture reaches 98% accuracy level on unseen data when trained with just 20 000 samples, compared to 85% accuracy even with 100 000 samples for the typical network of choice in mechanics research. We show that, with relatively high data efficiency, CNNs have the capability to generalize well and automatically learn deep symmetry operations, easily extending to higher dimensions and our 2D case. Most importantly, we show how CNNs can naturally represent mechanical material tensors, with its convolution kernels serving as local receptive fields, which is a natural representation of mechanical response. Strategies proposed are applicable to other mechanics' problems and may, in the future, be used to sidestep cumbersome algorithms with purely data-driven approaches based upon modern deep architectures.     

图像处理与识别在果蔬成熟度监测中的研究及应用

目的 分析了果蔬成熟度自动监测对发展智慧农业的重要意义,对图像处理与识别技术在监测果蔬成熟度领域的研究与应用现状进行综述、总结与展望,以期为我国发展果蔬成熟度在线或自动检测识别技术提供参考。方法 对图像处理与识别在监测果蔬成熟度中的原理、优势进行分析,对特征提取、深度学习中的神经网络在该领域中的应用研究进展进行综述。结果 采用以图像处理和识别为核心的计算机视觉检测技术对果蔬的颜色、纹理等外部特征进行成熟度检测具有优势,结合神经网络对果蔬成熟度进行检测的识别率高,可在采摘、运输等场景对果蔬成熟度进行监测。结论 图像处理与识别技术在果蔬成熟度监测领域有望得到突破,将催生更多新的应用场景。     

3D Head Pose Estimation through Facial Features and Deep Convolutional Neural Networks

Face image analysis is one among several important cues in computer vision. Over the last five decades, methods for face analysis have received immense attention due to large scale applications in various face analysis tasks. Face parsing strongly benefits various human face image analysis tasks inducing face pose estimation. In this paper we propose a 3D head pose estimation framework developed through a prior end to end deep face parsing model. We have developed an end to end face parts segmentation framework through deep convolutional neural networks (DCNNs). For training a deep face parts parsing model, we label face images for seven different classes, including eyes, brows, nose, hair, mouth, skin, and back. We extract features from gray scale images by using DCNNs. We train a classifier using the extracted features. We use the probabilistic classification method to produce gray scale images in the form of probability maps for each dense semantic class. We use a next stage of DCNNs and extract features from grayscale images created as probability maps during the segmentation phase. We assess the performance of our newly proposed model on four standard head pose datasets, including Pointing’04, Annotated Facial Landmarks in the Wild (AFLW), Boston University (BU), and ICT-3DHP, obtaining superior results as compared to previous results.     

Perceptual control and feedback control in the analysis of complex tasks

Despite its extensive and successful use in the human factors specialist's work, there remain challenges for the development of task analysis. One such challenge is posed by the need to capture the features of the dynamic, complex tasks that take place in modern socio-technical systems. In this paper, we discuss the theoretical and practical implications of using perceptual control theory (PCT) as a theoretical grounding for task analysis. In particular, we describe the ability of PCT to combine the notion of perceptual control (which is similar to the assumptions underlying ecological design) with that of feedback control (which is fundamental to some traditional task analysis approaches). We describe some of the current PCT-based task analysis methods before introducing a new method that aims to integrate PCT concepts into hierarchical task analysis. Finally, we demonstrate how this method might be applied to a real-world dynamic control task.     

Progress in Neural NLP: Modeling,Learning, and Reasoning

Natural language processing (NLP) is a subfield of artificial intelligence that focuses on enabling computers to understand and process human languages. In the last five years, we have witnessed the rapid development of NLP in tasks such as machine translation, question-answering, and machine reading comprehension based on deep learning and an enormous volume of annotated and unannotated data. In this paper, we will review the latest progress in the neural network-based NLP framework (neural NLP) from three perspectives: modeling, learning, and reasoning. In the modeling section, we will describe several fundamental neural network-based modeling paradigms, such as word embedding, sentence embedding, and sequence-to-sequence modeling, which are widely used in modern NLP engines. In the learning section, we will introduce widely used learning methods for NLP models, including supervised, semi-supervised, and unsupervised learning; multitask learning; transfer learning; and active learning. We view reasoning as a new and exciting direction for neural NLP, but it has yet to be well addressed. In the reasoning section, we will review reasoning mechanisms, including the knowledge, existing non-neural inference methods, and new neural inference methods. We emphasize the importance of reasoning in this paper because it is important for building interpretable and knowledge-driven neural NLP models to handle complex tasks. At the end of this paper, we will briefly outline our thoughts on the future directions of neural NLP.