Skin lesion classification based on the VGG-16 fusion residual structure

The analysis of skin lesion images is challenging due to the high interclass similarity and intraclass variance. Therefore, improving the ability to automatically classify based on skin lesion images is necessary to help physicians classify skin lesions. We propose a network model based on the Visual Geometry Group Network (VGG-16) fusion residual structure for the multiclass classification of skin lesions. based on the VGG-16 network, we simplify and improve the network structure by adding a preprocessing layer (CBRM layer) and fusing the residual structure. We also use a hair removal algorithm and perform six data augmentation operations on a small number of skin lesion images to balance the total number of the seven skin lesions in the dataset. The model was evaluated on the ISIC2018 dataset. Experiments have shown that our network model achieves good classification performance, with a test accuracy rate of 88.14% and a macroaverage of 98%.     

Bio-inspired Hybrid Feature Selection Model for Intrusion Detection

Intrusion detection is a serious and complex problem. Undoubtedly due to a large number of attacks around the world, the concept of intrusion detection has become very important. This research proposes a multilayer bio-inspired feature selection model for intrusion detection using an optimized genetic algorithm. Furthermore, the proposed multilayer model consists of two layers (layers 1 and 2). At layer 1, three algorithms are used for the feature selection. The algorithms used are Particle Swarm Optimization (PSO), Grey Wolf Optimization (GWO), and Firefly Optimization Algorithm (FFA). At the end of layer 1, a priority value will be assigned for each feature set. At layer 2 of the proposed model, the Optimized Genetic Algorithm (GA) is used to select one feature set based on the priority value. Modifications are done on standard GA to perform optimization and to fit the proposed model. The Optimized GA is used in the training phase to assign a priority value for each feature set. Also, the priority values are categorized into three categories: high, medium, and low. Besides, the Optimized GA is used in the testing phase to select a feature set based on its priority. The feature set with a high priority will be given a high priority to be selected. At the end of phase 2, an update for feature set priority may occur based on the selected features priority and the calculated F-Measures. The proposed model can learn and modify feature sets priority, which will be reflected in selecting features. For evaluation purposes, two well-known datasets are used in these experiments. The first dataset is UNSW-NB15, the other dataset is the NSL-KDD. Several evaluation criteria are used, such as precision, recall, and F-Measure. The experiments in this research suggest that the proposed model has a powerful and promising mechanism for the intrusion detection system.     

Autonomous Parking-Lots Detection with Multi-Sensor Data Fusion Using Machine Deep Learning Techniques

The rapid development and progress in deep machine-learning techniques have become a key factor in solving the future challenges of humanity. Vision-based target detection and object classification have been improved due to the development of deep learning algorithms. Data fusion in autonomous driving is a fact and a prerequisite task of data preprocessing from multi-sensors that provide a precise, well-engineered, and complete detection of objects, scene or events. The target of the current study is to develop an in-vehicle information system to prevent or at least mitigate traffic issues related to parking detection and traffic congestion detection. In this study we examined to solve these problems described by (1) extracting region-of-interest in the images (2) vehicle detection based on instance segmentation, and (3) building deep learning model based on the key features obtained from input parking images. We build a deep machine learning algorithm that enables collecting real video-camera feeds from vision sensors and predicting free parking spaces. Image augmentation techniques were performed using edge detection, cropping, refined by rotating, thresholding, resizing, or color augment to predict the region of bounding boxes. A deep convolutional neural network F-MTCNN model is proposed that simultaneously capable for compiling, training, validating and testing on parking video frames through video-camera. The results of proposed model employing on publicly available PK-Lot parking dataset and the optimized model achieved a relatively higher accuracy 97.6% than previous reported methodologies. Moreover, this article presents mathematical and simulation results using state-of-the-art deep learning technologies for smart parking space detection. The results are verified using Python, TensorFlow, OpenCV computer simulation frameworks.     

OCT skin image enhancement through attenuation compensation

The enhancement of optical coherence tomography (OCT) skin images can help dermatologists investigate the morphologic information of the images more effectively. In this paper, we propose an enhancement algorithm with the stages that includes speckle reduction, skin layer detection, and attenuation compensation. A weighted median filter is designed to reduce the level of speckle while preserving the contrast. A novel skin layer detection technique is then applied to outline the main skin layers: stratum corneum, epidermis, and dermis. The skin layer detection algorithm does not make any assumption about the structure of the skin. A model of the light attenuation is then used to estimate the attenuation coefficient of the stratum corneum, epidermis, and dermis layers. The performance of the algorithm has been evaluated qualitatively based on visual evaluation and quantitatively using two no-reference quality metrics: signal-to-noise ratio and contrast-to-noise ratio. The enhancement algorithm is tested on 35 different skin OCT images, which show significant improvements in the quality of the images, especially in the structures at deeper levels.     

小样本药用玻璃瓶缺陷生成方法及应用

目的 为了克服药包玻璃瓶缺陷样本不足带来的缺陷检测模型精度不准的问题,提出改进StyleGAN2-ADA的缺陷样本生成方法,提升模型鲁棒性。方法 首先,基于StyleGAN2-ADA算法,在无缺陷图像集上训练网络模型并作为骨干。其次,在骨干网络上添加缺陷感知残差块,生成缺陷掩码,在少量的缺陷图像数据集上训练网络模型操纵掩码区域的特征,模拟缺陷图像生成过程,合成缺陷图像。最后,采用YOLOv7检测网络验证该样本生成方法的效果。结果 实验结果表明,该方法在大量正常图像和少量缺陷图像基础上生成逼真且多样性的缺陷图像,应用该缺陷样本合成方法丰富数据集后,西林瓶缺陷检测平均准确率（mAP）达到97.3%,较原始数据集合和StyleGAN2-ADA算法分别提高了33.1%和4.1%。结论 该图像生成方法可以在少量缺陷样本下生成高质量的缺陷图像,优化不均衡数据集,增强模型训练的稳定性,提高药用玻璃包装产品的质量和合格率。     

基于胶囊网络的汉字笔迹鉴定算法

由于采集脱机汉字手写样本时忽略了书写人的心理和生理等因素对书写活动的影响,因而传统笔迹鉴定算法的泛化能力较低。针对上述问题,提出基于胶囊网络的汉字笔迹鉴定算法,并构建了跟踪采集数据集以模拟复杂背景下产生的汉字。胶囊网络构建活动向量表示特定类型的实例化参数,通过动态路由算法将活动向量路由到下一层相应的胶囊中,使下一层胶囊得到更清晰的输入信号。分别采用5种算法对HWDB1.1数据集和跟踪采集数据集进行了测试,实验结果表明：本文算法的分类准确率比其他4种算法的都高,HWDB1.1数据集和跟踪采集数据集中算法的分类准确率分别为95.82%, 94.39%;本文算法具有较强的泛化性能,对训练样本数的依赖程度较低,弥补了卷积神经网络池化层的信息丢失缺陷。     

Few-Shot Learning with Generative Adversarial Networks Based on WOA13 Data

In recent years, extreme weather events accompanying the global warming have occurred frequently, which brought significant impact on national economic and social development. The ocean is an important member of the climate system and plays an important role in the occurrence of climate anomalies. With continuous improvement of sensor technology, we use sensors to acquire the ocean data for the study on resource detection and disaster prevention, etc. However, the data acquired by the sensor is not enough to be used directly by researchers, so we use the Generative Adversarial Network (GAN) to enhance the ocean data. We use GAN to process WOA13 dataset and use ResNet to determine if there is a thermocline layer in a sea area. We compare the classification results of the enhanced datasets of different orders of magnitude with the classification results of the original datasets. The experimental result shows that the dataset processed by GAN has a higher accuracy. GAN has a certain enhancement effect to marine data. Gan increased the accuracy of the WOA dataset from 0.91 to 0.93. At the same time, the experimental results also show that too much data cannot continue to enhance the accuracy of WOA in ResNet.     

内填脱硫石膏砌块墙体的新型装配式钢框架抗震性能研究

为分析脱硫石膏砌块填充墙体对新型装配式钢框架抗震性能的影响,开展了2榀1/2缩尺的单跨2层装配式钢框架的拟静力试验研究。通过与纯钢框架试件的对比,得到了脱硫石膏砌块填充墙体的破坏模式及其对钢框架抗震性能的影响。结果表明:采用新型装配式节点的钢框架具有良好的抗震性能;脱硫石膏砌块填充墙体不仅显著提高了框架的初始抗侧刚度,而且对框架的延性、耗能均有不同程度的提高。采用数值建模的方法对试验进行模拟,其结果与试验结果吻合良好,并针对试件的轴压比、高跨比和砌块厚度进行变参分析,得出增大试件的轴压比对试件的延性有一定的削弱作用,建议轴压比取值小于0.4;过大或过小的高跨比对结构均不利;改变砌块厚度对结构初始刚度影响较大,但对结构的承载能力影响较小。     

An Efficient Crossing-Line Crowd Counting Algorithm with Two-Stage Detection

Crowd counting is a challenging task in crowded scenes due to heavy occlusions, appearance variations and perspective distortions. Current crowd counting methods typically operate on an image patch level with overlaps, then sum over the patches to get the final count. In this paper we describe a real-time pedestrian counting framework based on a two-stage human detection algorithm. Existing works with overhead cameras is mainly based on visual tracking, and their robustness is rather limited. On the other hand, some works, which focus on improving the performances, are too complicated to be realistic. By adopting a line sampling process, a temporal slice image can be obtained for pedestrian counting without the need for visual tracking. Only ten low level features are extracted from the input image to establish a feature vector. As a result, our algorithm is more efficient and accurate than existing methods. Pedestrians in the temporal slice image are then located by the two-stage detection algorithm, which is largely based on support vector machine and affinity propagation clustering. Moreover, a novel algorithm is proposed to determine the moving directions of pedestrians by comparing the centers of them in two temporal slice images. Extensive experiments reveal that our system achieves satisfaction performances in terms of both robustness and efficiency.     

Adaptive Binary Coding for Scene Classification Based on Convolutional Networks

With the rapid development of computer technology, millions of images are produced everyday by different sources. How to efficiently process these images and accurately discern the scene in them becomes an important but tough task. In this paper, we propose a novel supervised learning framework based on proposed adaptive binary coding for scene classification. Specifically, we first extract some high-level features of images under consideration based on available models trained on public datasets. Then, we further design a binary encoding method called one-hot encoding to make the feature representation more efficient. Benefiting from the proposed adaptive binary coding, our method is free of time to train or fine-tune the deep network and can effectively handle different applications. Experimental results on three public datasets, i.e., UIUC sports event dataset, MIT Indoor dataset, and UC Merced dataset in terms of three different classifiers, demonstrate that our method is superior to the state-of-the-art methods with large margins.     

Biomedical Event Extraction Using a New Error Detection Learning Approach Based on Neural Network

Supervised machine learning approaches are effective in text mining, but their success relies heavily on manually annotated corpora. However, there are limited numbers of annotated biomedical event corpora, and the available datasets contain insufficient examples for training classifiers; the common cure is to seek large amounts of training samples from unlabeled data, but such data sets often contain many mislabeled samples, which will degrade the performance of classifiers. Therefore, this study proposes a novel error data detection approach suitable for reducing noise in unlabeled biomedical event data. First, we construct the mislabeled dataset through error data analysis with the development dataset. The sample pairs’ vector representations are then obtained by the means of sequence patterns and the joint model of convolutional neural network and long short-term memory recurrent neural network. Following this, the sample identification strategy is proposed, using error detection based on pair representation for unlabeled data. With the latter, the selected samples are added to enrich the training dataset and improve the classification performance. In the BioNLP Shared Task GENIA, the experiments results indicate that the proposed approach is competent in extract the biomedical event from biomedical literature. Our approach can effectively filter some noisy examples and build a satisfactory prediction model.     

AED-Net: An Abnormal Event Detection Network

It has long been a challenging task to detect an anomaly in a crowded scene. In this paper, a self-supervised framework called the abnormal event detection network (AED-Net), which is composed of a principal component analysis network (PCAnet) and kernel principal component analysis (kPCA), is proposed to address this problem. Using surveillance video sequences of different scenes as raw data, the PCAnet is trained to extract high-level semantics of the crowd’s situation. Next, kPCA, a one-class classifier, is trained to identify anomalies within the scene. In contrast to some prevailing deep learning methods, this framework is completely self-supervised because it utilizes only video sequences of a normal situation. Experiments in global and local abnormal event detection are carried out on Monitoring Human Activity dataset from University of Minnesota (UMN dataset) and Anomaly Detection dataset from University of California, San Diego (UCSD dataset), and competitive results that yield a better equal error rate (EER) and area under curve (AUC) than other state-of-the-art methods are observed. Furthermore, by adding a local response normalization (LRN) layer, we propose an improvement to the original AED-Net. The results demonstrate that this proposed version performs better by promoting the framework’s generalization capacity.     

Adaptive target detection with a polarization-sensitive optical system

We developed an adaptive polarimetric target detector (APTD) to determine the optimum combination strategy for a multichannel polarization-sensitive optical system. The proposed algorithm is based on scene-derived polarization properties of the target and background, and it seeks to find an optimum multichannel combination of linear polarizing filters that maximizes the signal-to-clutter ratio (SCR) in intensity and Stokes parameter images. The algorithm is validated by performing RX anomaly detection and a generalized likelihood ratio test on both synthetic and real imagery. The experimental results are analyzed through calculated SCR and receiver operating characteristic curves. Compared with several conventional operation methods, we find that better target detection performance is achieved with the APTD algorithm.     

改进YOLOv3算法用于铝型材表面缺陷检测

针对目前铝型材表面缺陷检测存在的准确率、检测效率较低等问题,提出了一种基于改进的YOLOv3铝型材表面缺陷检测方法。首先通过k-均值聚类算法对采集到的数据集进行聚类分析,选取尺寸最优的目标候选框;考虑到铝型材表面缺陷较大,对YOLOv3的网络层级结构进行调整,并将目标检测层之前的6个CBL单元改成4个,再补充2个残差单元,以提高特征的复用。将提出方法用于铝型材表面缺陷检测,并与经典的卷积网络Faster-RCNN和SSD方法进行比较,实验结果表明,采用提出的算法准确率达到97%,检测速度达到47帧/s,明显优于经典的卷积网络Faster-RCNN和SSD,适于在有高精度快速性要求的铝型材表面缺陷检测中推广应用。     

A facile method for fabrication of large area graphene nanostructures

In this study, we introduce a novel method to produce large area interconnected graphene nanostructures. A single layer CVD (Chemical Vapor Deposition) grown graphene was nanostructured by employing dewetted Ni thin film as an etching mask for the underlying graphene. As a result, a network of graphene nanostructures with irregular shapes and widths down to 10 nm is obtained. The FET (field effect transistor) devices fabricated employing the nanostructured graphene as channel material exhibit increased on/off current ratio compared to pristine graphene indicating a slight band gap opening due to the quantum confinement effect in such narrow graphene nanostructures. This technique can be useful for the large scale fabrication of graphene based electronic devices such as FETs and sensors.     

Monte Carlo study for physiological interference reduction in near-infrared spectroscopy based on empirical mode decomposition

Near-infrared spectroscopy (NIRS) can be used as the basis of non-invasive neuroimaging that may allow the measurement of haemodynamic changes in the human brain evoked by applied stimuli. Since this technique is very sensitive, physiological interference arising from the cardiac cycle and breathing can significantly affect the signal quality. Such interference is difficult to remove by conventional techniques because it occurs not only in the extracerebral layer but also in the brain tissue itself. Previous work on this problem employing temporal filtering, spatial filtering, and adaptive filtering have exhibited good performance for recovering brain activity data in evoked response studies. However, in this study, we present a time-frequency adaptive method for physiological interference reduction based on the combination of empirical mode decomposition (EMD) and Hilbert spectral analysis (HSA). Monte Carlo simulations based on a five-layered slab model of a human adult head were implemented to evaluate our methodology. We applied an EMD algorithm to decompose the NIRS time series derived from Monte Carlo simulations into a series of intrinsic mode functions (IMFs). In order to identify the IMFs associated with symmetric interference, the extracted components were then Hilbert transformed from which the instantaneous frequencies could be acquired. By reconstructing the NIRS signal by properly selecting IMFs, we determined that the evoked brain response is effectively filtered out with even higher signal-to-noise ratio (SNR). The results obtained demonstrated that EMD, combined with HSA, can effectively separate, identify and remove the contamination from the evoked brain response obtained with NIRS using a simple single source–detector pair.     

Multi-Scale Attention-Based Deep Neural Network for Brain Disease Diagnosis

Whole brain functional connectivity (FC) patterns obtained from resting-state functional magnetic resonance imaging (rs-fMRI) have been widely used in the diagnosis of brain disorders such as autism spectrum disorder (ASD). Recently, an increasing number of studies have focused on employing deep learning techniques to analyze FC patterns for brain disease classification. However, the high dimensionality of the FC features and the interpretation of deep learning results are issues that need to be addressed in the FC-based brain disease classification. In this paper, we proposed a multi-scale attention-based deep neural network (MSA-DNN) model to classify FC patterns for the ASD diagnosis. The model was implemented by adding a flexible multi-scale attention (MSA) module to the auto-encoder based backbone DNN, which can extract multi-scale features of the FC patterns and change the level of attention for different FCs by continuous learning. Our model will reinforce the weights of important FC features while suppress the unimportant FCs to ensure the sparsity of the model weights and enhance the model interpretability. We performed systematic experiments on the large multi-sites ASD dataset with both ten-fold and leave-one-site-out cross-validations. Results showed that our model outperformed classical methods in brain disease classification and revealed robust inter-site prediction performance. We also localized important FC features and brain regions associated with ASD classification. Overall, our study further promotes the biomarker detection and computer-aided classification for ASD diagnosis, and the proposed MSA module is flexible and easy to implement in other classification networks.     

Using Object Detection Network for Malware Detection and Identification in Network Traffic Packets

In recent years, the number of exposed vulnerabilities has grown rapidly and more and more attacks occurred to intrude on the target computers using these vulnerabilities such as different malware. Malware detection has attracted more attention and still faces severe challenges. As malware detection based traditional machine learning relies on exports’ experience to design efficient features to distinguish different malware, it causes bottleneck on feature engineer and is also time-consuming to find efficient features. Due to its promising ability in automatically proposing and selecting significant features, deep learning has gradually become a research hotspot. In this paper, aiming to detect the malicious payload and identify their categories with high accuracy, we proposed a packet-based malicious payload detection and identification algorithm based on object detection deep learning network. A dataset of malicious payload on code execution vulnerability has been constructed under the Metasploit framework and used to evaluate the performance of the proposed malware detection and identification algorithm. The experimental results demonstrated that the proposed object detection network can efficiently find and identify malicious payloads with high accuracy.     

基于改进YOLOv4的焊接件表面缺陷检测算法

目的 针对真实复杂的工业场景下焊接件表面缺陷检测精度低、速度慢和图像噪声大等问题,提出一种基于卷积神经网络的改进YOLOv4焊接件表面缺陷检测算法。方法 该模型基于YOLOv4算法,首先,考虑到存储和计算资源的限制,使用了轻量级网络GhostNet替换YOLOv4的主干特征提取网络(Backbone)CSPDarknet53;其次,在GhostNet网络结构中嵌入改进的通道注意力机制,能够提高模型的学习能力且减少参数量;最后,引入K–means++聚类算法对焊接件表面缺陷数据集中待检测的标注框宽高进行聚类,使网络模型更容易检测到样本中的缺陷。结果 实验结果表明,改进后的YOLOv4算法平均精度(mean Average Precision,mAP)为91.07%,检测速度达到48.11帧/s,模型尺寸为43.2 MB,比原始YOLOv4算法平均精度提升了4.61%,检测速度提高了26.59帧/s,模型尺寸缩减了82.37%。结论 所提模型提高了焊接件表面缺陷检测的精度和速度,在工业表面缺陷检测中具有现实意义。     

A Model for Helmet-Wearing Detection of Non-Motor Drivers Based on YOLOv5s

Detecting non-motor drivers’ helmets has significant implications for traffic control. Currently, most helmet detection methods are susceptible to the complex background and need more accuracy and better robustness of small object detection, which are unsuitable for practical application scenarios. Therefore, this paper proposes a new helmet-wearing detection algorithm based on the You Only Look Once version 5 (YOLOv5). First, the Dilated convolution In Coordinate Attention (DICA) layer is added to the backbone network. DICA combines the coordinated attention mechanism with atrous convolution to replace the original convolution layer, which can increase the perceptual field of the network to get more contextual information. Also, it can reduce the network’s learning of unnecessary features in the background and get attention to small objects. Second, the Rebuild Bidirectional Feature Pyramid Network (Re-BiFPN) is used as a feature extraction network. Re-BiFPN uses cross-scale feature fusion to combine the semantic information features at the high level with the spatial information features at the bottom level, which facilitates the model to learn object features at different scales. Verified on the proposed “Helmet Wearing dataset for Non-motor Drivers (HWND),” the results show that the proposed model is superior to the current detection algorithms, with the mean average precision (mAP) of 94.3% under complex background.