Deep Transfer Learning Models for Mobile-Based Ocular Disorder Identification on Retinal Images

Mobile technology is developing significantly.Mobile phone technologies have been integrated into the healthcare industry to help medical practitioners.Typically,computer vision models focus on image detection and classification issues.MobileNetV2 is a computer vision model that performs well on mobile devices,but it requires cloud services to process biometric image information and provide predictions to users.This leads to increased latency.Processing biometrics image datasets on mobile devices will make the prediction faster,but mobiles are resource-restricted devices in terms of storage,power,and computational speed.Hence,a model that is small in size,efficient,and has good prediction quality for biometrics image classification problems is required.Quantizing pre-trained CNN(PCNN)MobileNetV2 architecture combined with a Support Vector Machine(SVM)compacts the model representation and reduces the computational cost and memory requirement.This proposed novel approach combines quantized pre-trained CNN(PCNN)MobileNetV2 architecture with a Support Vector Machine(SVM)to represent models efficiently with low computational cost and memory.Our contributions include evaluating three CNN models for ocular disease identification in transfer learning and deep feature plus SVM approaches,showing the superiority of deep features from MobileNetV2 and SVM classification models,comparing traditional methods,exploring six ocular diseases and normal classification with 20,111 images postdata augmentation,and reducing the number of trainable models.The model is trained on ocular disorder retinal fundus image datasets according to the severity of six age-related macular degeneration(AMD),one of the most common eye illnesses,Cataract,Diabetes,Glaucoma,Hypertension,andMyopia with one class Normal.From the experiment outcomes,it is observed that the suggested MobileNetV2-SVM model size is compressed.The testing accuracy for MobileNetV2-SVM,InceptionV3,and MobileNetV2 is 90.11%,86.88%,and 89.76%respectively while MobileNetV2-SVM,InceptionV3,and MobileNetV2 accuracy are observed to be 92.59%,83.38%,and 90.16%,respectively.The proposed novel technique can be used to classify all biometric medical image datasets on mobile devices.

Research on Fall Detection System Based on Commercial Wi-Fi Devices

Falls are a major cause of disability and even death in the elderly,and fall detection can effectively reduce the damage.Compared with cameras and wearable sensors,Wi-Fi devices can protect user privacy and are inexpensive and easy to deploy.Wi-Fi devices sense user activity by analyzing the channel state information(CSI)of the received signal,which makes fall detection possible.We propose a fall detection system based on commercial Wi-Fi devices which achieves good performance.In the feature extraction stage,we select the discrete wavelet transform(DWT)spectrum as the feature for activity classification,which can balance the temporal and spatial resolution.In the feature classification stage,we design a deep learning model based on convolutional neural networks,which has better performance compared with other traditional machine learning models.Experimental results show our work achieves a false alarm rate of 4.8%and a missed alarm rate of 1.9%.

ANN-Based Identification of Steady-State Behavior Parameters of Composite Power Semiconductor Device Model

The paper describes the application of an ANN based approach to the identification of the parameters relevant to the steady state behavior of composite power electronic device models of circuit simulation software. The identification of model parameters of IGBT in PSPICE using BP neural network is illustrated.

Identifying Honeypots from ICS Devices Using Lightweight Fuzzy Testing

The security issues of industrial control systems(ICSs)have become increasingly prevalent.As an important part of ICS security,honeypots and anti-honeypots have become the focus of offensive and defensive confrontation.However,research on ICS honeypots still lacks breakthroughs,and it is difficult to simulate real ICS devices perfectly.In this paper,we studied ICS honeypots to identify and address their weaknesses.First,an intelligent honeypot identification framework is proposed,based on which feature data type requirements and feature data acquisition for honeypot identification is studied.Inspired by vulnerability mining,we propose a feature acquisition approach based on lightweight fuzz testing,which utilizes the differences in error handling between the ICS device and the ICS honeypot.By combining the proposed method with common feature acquisition approaches,the integrated feature data can be obtained.The experimental results show that the feature data acquired is effective for honeypot identification.

A Material Identification Approach Based on Wi-Fi Signal

Material identification is a technology that can help to identify the type of target material.Existing approaches depend on expensive instruments,complicated pre-treatments and professional users.It is difficult to find a substantial yet effective material identification method to meet the daily use demands.In this paper,we introduce a Wi-Fi-signal based material identification approach by measuring the amplitude ratio and phase difference as the key features in the material classifier,which can significantly reduce the cost and guarantee a high level accuracy.In practical measurement of WiFi based material identification,these two features are commonly interrupted by the software/hardware noise of the channel state information(CSI).To eliminate the inherent noise of CSI,we design a denoising method based on the antenna array of the commercial off-the-shelf(COTS)Wi-Fi device.After that,the amplitude ratios and phase differences can be more stably utilized to classify the materials.We implement our system and evaluate its ability to identify materials in indoor environment.The result shows that our system can identify 10 commonly seen liquids with an average accuracy of 98.8%.It can also identify similar liquids with an overall accuracy higher than 95%,such as various concentrations of salt water.

A Framework for Multi-Hop Ad-Hoc Networking over Wi-Fi Direct with Android Smart Devices

The wide diffusion of mobile devices that natively support ad hoc communication technologies has led to several protocols for enabling and optimizing Mobile Ad Hoc Networks (MANETs). Nevertheless, the actual utilization of MANETs in real life seems limited due to the lack of protocols for the automatic creation and evolution of ad hoc networks. Recently, a novel P2P protocol named Wi-Fi Direct has been proposed and standardized by the Wi-Fi Alliance to facilitate nearby devices’ interconnection. Wi-Fi Direct provides high-performance direct communication among devices, includes different energy management mechanisms, and is now available in most Android mobile devices. However, the current implementation of Wi-Fi Direct on Android has several limitations, making the Wi-Fi Direct network only be a one-hop ad-hoc network. This paper aims to develop a new framework for multi-hop ad hoc networking using Wi-Fi Direct in Android smart devices. The framework includes a connection establishment protocol and a group management protocol. Simulations validate the proposed framework on the OMNeT++ simulator. We analyzed the framework by varying transmission range, number of hops, and buffer size. The results indicate that the framework provides an eventual 100% packet delivery for different transmission ranges and hop count values. The buffer size has enough space for all packets. However, as buffer size decreases, the packet delivery decreases proportionally.

科胜讯和Ozmo Devices推出针对WI-FI~外设的音频参考设计

为影像、音频、嵌入式调制解调器和视频应用提供创新半导体解决方案的领先供应商科胜讯系统公司和Wi-Fi个人局域网（Wi—Fi PAN）领先厂商Ozmo Devices共同宣布联合开发扬声器和免提电话参考设计。

科胜讯和Ozmo Devices针对WI-FI推出外设的音频参考设计

科胜讯系统公司和Ozmo Devices共同宣布联合开发扬声器和免提电话参考设计。 新的解决方案基于科胜讯创新的CX2070X音频系统级芯片（SoC）产品系列和OZM01000超低功耗Wi-Fi PAN SoC。科胜讯的CX2070X SoC采用专利的技术创新，

MATERIAL SURFACE THERMAL PROPERTY IDENTIFICATION USING HEAT FLUX TACTILE SENSOR

Eased on the mechanism of temperature tactile sensing of human finger,a heat flux tactile sensor com- posed of a thermostat module and a heat flux sensor is designed to identify material thermal properties. The ther- mostat module maintains the sensor temperature invariable, and the heat flux sensor（Peltier device） detects the heat flux temperature difference between the thermostat module and the object surface. Two different modes of the heat flux tactile sensor are proposed, and they are simulated and experimented for different material objects. The results indicate that the heat flux tactile sensor can effectively identify different thermal properties.

An accurate identification method for network devices based on spatial attention mechanism

With the metaverse being the development direction of the next generation Internet,the popularity of intelligent devices,and the maturity of various emerging technologies,more and more intelligent devices try to connect to the Internet,which poses a major threat to the management and security protection of network equipment.At present,the mainstream method of network equipment identification in the metaverse is to obtain the network traffic data generated in the process of device communication,extract the device features through analysis and processing,and identify the device based on a variety of learning algorithms.Such methods often require manual participation,and it is difficult to capture the small differences between similar devices,leading to identification errors.Therefore,we propose a deep learning device recognition method based on a spatial attention mechanism.Firstly,we extract the required feature fields from the acquired network traffic data.Then,we normalize the data and convert it into grayscale images.After that,we add a spatial attention mechanism to CNN and MLP respectively to increase the difference between similar network devices and further improve the recognition accuracy.Finally,we identify devices based on the deep learning model.A large number of experiments were carried out on 31 types of network devices such as web cameras,wireless routers,and smartwatches.The results show that the accuracy of the proposed recognition method based on the spatial attention mechanism is increased by 0.8%and 2.0%,respectively,compared with the recognition method based only on the deep learning model under the CNN and MLP models.The method proposed in this paper is significantly superior to the existing method of device-type recognition based only on a deep learning model.

基于主动探测的Web容器探测识别方法

随着工业互联网的飞速发展,各类Web容器的广泛使用呈现不断增长的趋势,然而,这也使得Web容器资产管理的问题变得更加复杂.随之而来的是诸多网络安全风险和潜在隐患,对于这些挑战,提升网络安全防御水平显得尤为迫切.为了解决这一问题,本文引入了一种新的基于主动探测的Web容器探测识别方法.在探测阶段,采用了一种先进的Web容器探针构建方法,通过此方法构建了Web容器探针.这个探针在识别阶段发挥关键作用,借助一种基于负载内容的Web容器识别方法,通过协议解码技术,实现了对Web容器的高度准确的识别.通过结合这两种先进的识别方法,成功识别了4种不同类型的Web容器,并且提升了精度,能够精确地区分这些Web容器的各个版本,总计实现了10个版本的准确识别.通过这种先进的主动探测方法,企业可以更好地了解和管理其Web容器资产,降低网络安全风险,并确保网络系统的稳定性和安全性.

基于UDI技术的单件器械追溯系统建设与应用探讨

为提高单件器械在回收、清洗、核包、灭菌及使用等环节的追溯精准度,提高贵重器械的利用率和成本管控水平,医院在消毒供应中心物品追溯系统信息化平台上,开发了“基于UDI技术的消毒供应中心单件器械追溯系统”。通过在单件器械上刻印UDI,在回收、器械组配两个环节扫描UDI码,记录单件器械的使用次数、单包清洗灭菌次数和单包使用明细,有效减少不必要的浪费,降低清洗消毒成本和器械采购开支,为手术质量管理提供支持。

物联网设备识别及异常检测研究综述

随着物联网技术的发展,物联网设备广泛应用于生产和生活的各个领域,但也为设备资产管理和安全管理带来了严峻的挑战.首先,由于物联网设备类型和接入方式的多样性,网络管理员通常难以得知网络中的物联网设备类型及运行状态.其次,物联网设备由于其计算、存储资源有限,难以部署传统防御措施,正逐渐成为网络攻击的焦点.因此,通过设备识别了解网络中的物联网设备并基于设备识别结果进行异常检测,以保证其正常运行尤为重要.近几年来,学术界围绕上述问题开展了大量的研究.系统地梳理物联网设备识别和异常检测方面的相关工作.在设备识别方面,根据是否向网络中发送数据包,现有研究可分为被动识别方法和主动识别方法.针对被动识别方法按照识别方法、识别粒度和应用场景进行进一步的调研,针对主动识别方法按照识别方法、识别粒度和探测粒度进行进一步的调研.在异常检测方面,按照基于机器学习算法的检测方法和基于行为规范的规则匹配方法进行梳理.在此基础上,总结物联网设备识别和异常检测领域的研究挑战并展望其未来发展方向.

一种分布式会议管理系统的设计与实现

基于现代会议管理的需求,设计和实现一种C/S和B/S混合部署的会议管理系统。包括会议管理服务中心和若干个会议现场,会议管理服务中心包括数据服务器、应用服务器、Web服务器、通信网关和出口路由器;会议现场包括若干个便携式电脑、RFID(Radio Frequency Identification)读卡器、二维码阅读器、信息显示发布设备、现场WLAN设备及用户终端。使用RIA(Rich Internet Application)技术优化了B/S界面,应用RFID对会议过程中的细节进行监控,借助SAAS(Software as a Service)模式实现会议管理按需配置和快速部署。该系统显著提高了会议管理效率。

SPD助力手术室高值医用耗材精细化管理

对生产厂家、供应商、物流公司、设备物资科及手术室进行规范化管理,利用SPD医用耗材管理系统及智能感知柜管理系统,改变原有手术室高值医用耗材逆向物流的模式,建立院内手术室高值医用耗材的正向物流管理模式及追溯机制,以降低临床医用耗材管理工作强度,实现高值医用耗材可视化管理和精细化管理,为医院进行科学合理的宏观调控提供可靠准确的信息,在保证高值医用耗材使用快捷的同时,最大限度地保证手术室高值医用耗材在使用中的安全性及可追溯性。

低压直流固态断路器关键技术研究

相较于机械式断路器,固态断路器因其更快的分断速度和更高的寿命等优点被广泛关注。介绍了一种基于SiC器件的低压直流固态断路器拓扑和工作原理,分析其技术难点,包括低感封装及器件设计、隔离开关技术和快速故障识别技术。研制了一种±375 V直流固态断路器样机,同时进行了故障电流关断测试和温升测试。结果表明所设计直流固态断路器能够在百微秒内实现1 kA短路电流分断,并具备快速故障识别、保护曲线可编程、外部通信与远程控制等功能。

武器站伺服传动装置在线辨识与自适应控制方法

针对武器站俯仰轴伺服传动装置在武器载荷变化时控制精度和安全性下降的问题,提出基于多参数在线辨识的自适应复合控制方法。基于符号函数的线性化方法,构建包含传动间隙、电机和负载摩擦非线性的伺服传动装置待辨识模型框架,推导出待辨识参数的显式迭代方程,通过带遗忘因子的递推增广最小二乘算法实现参数在线辨识。在此基础上,提出自适应比例积分控制器与自适应状态扩张卡尔曼滤波器相结合的复合控制方法。实验结果表明:参数辨识方法可实现伺服传动装置9个关键动力学参数的精确在线辨识,稳态辨识误差不超过10%。自适应复合控制方法将系统速度跟随残差均方根降低了28.24%,有效提升了俯仰轴在武器载荷变化时的控制精度和稳定裕度。

基于AUKF的可穿戴式设备用锂离子电池SOE在线估计方法

可穿戴式设备(wearable devices,WDs)体积小、工作时间长,在工业监测等领域应用越来越广泛。锂离子电池为WD上电子设备提供能量,其准确的能量状态(state of energy,SOE)在线估算对WDs的电源实时管理与延长设备寿命有重要影响。传统的基于模型的估算方法需要离线获取SOE与开路电压(open circuit voltage,OCV)的关系,实验时间长,不能适应实际工况变化,难以在线实施,本工作提出一种基于开路电压(opencircuit voltage,OCV)在线辨识的可穿戴式设备用锂离子电池SOE在线估算方法。首先基于锂离子电池的一阶RC模型,采用带遗忘因子的递推最小二乘法(forgetting factor recursive least squares,FFRLS)在线辨识电池OCV等参数。分析了WDs运行负载变化特征,构建了WD运行工况和参数辨识工况,并开展锂离子电池实验。结合WDs工作负载特性,研究了开路电压和端电压的关系,在线获得OCV与SOE的关系曲线。采用无迹卡尔曼滤波(adaptive unscented Kalman filter,AUKF)算法实现SOE的在线估计,与传统通过离线实验获得OCVSOE关系的方法进行了对比。研究结果表明,所提的SOE在线估算方法具有较好的精度,并在不同的SOE初始值时具有较好的鲁棒性。

含风电场的分数阶电力系统自适应同步控制

针对含风电场的分数阶电力系统的混沌振荡问题,基于自适应同步理论,提出了一种自适应同步控制方法。首先,建立了一个整数阶三维且含储能装置的电力系统模型并推广为分数阶,采用相图、时序图、分岔图等方法对含风电场的分数阶电力系统的动力学行为进行分析。其次,推导了自适应同步控制定理的证明。最后,通过含有待辨识参数且含风电场分数阶混沌电力系统与稳定状态的系统实现完全同步,间接地实现系统的混沌控制和系统的参数辨识。

可区分应力集中和缺陷的双线圈共磁芯式梯度测磁传感装置设计

针对复杂构件特殊位置熔焊缝无法全覆盖检测的工程问题,基于磁检测技术,设计了一种双线圈共磁芯式梯度测磁传感装置。以S06马氏体不锈钢加工的熔焊缝试块为实验对象,通过对比不同提离高度下焊接缺陷和应力集中处的磁信号检测结果,以及对传感装置的性能测试,确定装置参数。结果表明,在应力集中和焊接缺陷处,磁感应强度均随着传感器提离值的增大而减小,且在一定的提离高度下,缺陷磁异常特征消失,而应力集中处磁异常信号仍然存在,可据此来区分应力集中和焊接缺陷。射线检测结果验证了所设计传感装置在焊缝缺陷检测上的可行性与有效性。对于材料磁性和焊接工艺相似的熔焊缝构件,可采取同样的试验方法标定装置参数,以扩大所设计装置的适用性。