COVID TCL: A Joint Metric Loss Function for Diagnosing COVID-19 Patient in the Early and Incubation Period

Convolution Neural Networks (CNN) can quickly diagnose COVID-19 patients by analyzing computed tomography (CT) images of the lung, thereby effectively preventing the spread of COVID-19. However, the existing CNN-based COVID-19 diagnosis models do consider the problem that the lung images of COVID-19 patients in the early stage and incubation period are extremely similar to those of the non-COVID-19 population. Which reduces the model’s classification sensitivity, resulting in a higher probability of the model misdiagnosing COVID-19 patients as non-COVID-19 people. To solve the problem, this paper first attempts to apply triplet loss and center loss to the field of COVID-19 image classification, combining softmax loss to design a jointly supervised metric loss function COVID Triplet-Center Loss (COVID-TCL). Triplet loss can increase inter-class discreteness, and center loss can improve intra-class compactness. Therefore, COVID-TCL can help the CNN-based model to extract more discriminative features and strengthen the diagnostic capacity of COVID-19 patients in the early stage and incubation period. Meanwhile, we use the extreme gradient boosting (XGBoost) as a classifier to design a COVID-19 images classification model of CNN-XGBoost architecture, to further improve the CNN-based model’s classification effect and operation efficiency. The experiment shows that the classification accuracy of the model proposed in this paper is 97.41%, and the sensitivity is 97.61%, which is higher than the other 7 reference models. The COVID-TCL can effectively improve the classification sensitivity of the CNN-based model, the CNN-XGBoost architecture can further improve the CNN-based model’s classification effect.     

Hilbert resonator-based multiband tunable graphene metasurface polarizer for lower THz frequency

Journal of Computational Electronics - In this article, we propose different Hilbert shape-based tunable and multiband polarizers for the lower-terahertz frequency range. The tunability of this...     

Smart Prediction and Monitoring of Waste Disposal System Using IoT and Cloud for IoT Based Smart Cities

One of the prominent applications of Internet of Things (IoT) in this digital era is the development of smart cities. In IoT based smart cities, the smart objects (devices) are connected with each other via internet as a backbone. The sensed data by the smart objects are transmitted to the sink for further processing using multi hop communication. The smart cities use the analyzed data to improve their infrastructure, public utilities and they enhance their services by using the IoT technology for the betterment of livelihood of the common people. For IoT based smart cities, waste collection is a prominent issue for municipalities that aim to achieve a clean environment. With a boom in population in urban areas, an increasing amount of waste is generated. A major issue of waste management system is the poor process used in waste collection and segregation. Public bins begin to overflow for a long period before the process of cleaning starts, which is resulting in an accumulation of bacteria causing bad odors and spreading of diseases. In order to overcome this issue, in this paper an IoT based smart predication and monitoring of waste disposal system is proposed which utilizes off-the-shelf components that can be mounted to a bin of any size and measure fill levels. An Arduino microcontroller is employed in the proposed model to interface the infrared (IR), ultraviolet (UV), weight sensors, and a Global Positioning System (GPS) module is used to monitor the status of bins at predetermined intervals. The proposed system transmits the data using the cluster network to the master module which is connected to the backend via Wi-Fi. As data is collected, an intelligent neural network algorithm namely Long Short-Term Memory (LSTM) is used which will intelligently learn and predict the upcoming wastage from waste generation patterns. Moreover, the proposed system uses Firebase Cloud Messaging to notify the appropriate people when the bins were full and needed to be emptied. The Firebase Cloud Messaging (FCM) JavaScript Application Programming Interface (API) is used to send notification messages in web apps in browsers that provide service work support. Hence, the proposed system is useful to the society by providing facilities to the governments for enforcing stricter regulations for waste disposal. Additional features such as automated calibration of bin height, a dynamic web data dashboard as well as collation of data into a distributed real-time firebase database are also provided in the proposed system.

     

Performance Evaluation of Multilayer Clustering Network Using Distributed Energy Efficient Clustering with Enhanced Threshold Protocol

Wireless Personal Communications - In this research, pure deterministic system has been established by a new Distributed Energy Efficient Clustering Protocol with Enhanced Threshold (DEECET) by...     

Cyclic learning rate based HybridSN model for hyperspectral image classification

Classification of remotely sensed hyperspectral images (HSI) is a challenging task due to the presence of a large number of spectral bands and due to the less available data of remotely sensed HSI. The use of 3D-CNN and 2D-CNN layers to extract spectral and spatial features shows good test results. The recently introduced HybridSN model for the classification of remotely sensed hyperspectral images is the best to date compared to the other state-of-the-art models. But the test performance of the HybridSN model decreases significantly with the decrease in training data or number of training epochs. In this paper, we have considered cyclic learning for training of the HybridSN model, which shows a significant increase in the test performance of the HybridSN model with 10%, 20%, and 30% training data and limited number of training epochs. Further, we introduce a new cyclic function (ncf) whose training and test performance is comparable to the existing cyclic learning rate policies. More precisely, the proposed HybridSN(ncf ) model has higher average accuracy compared to HybridSN model by 19.47%, 1.81% and 8.33% for Indian Pines, Salinas Scene and University of Pavia datasets respectively in case of 10% training data and limited number of training epochs.

     

Microstructure and Property Characterization of Sintered Si3N4, SiC, and SiAlON

Commercially produced pressureless sintered Si₃N₄, SiC, and SiAlON were characterized with respect to density, phases present, bend strength, and oxidation resistance. The room-temperature bend strengths of sintered Si₃N₄, SiC, and SiAlON are comparable. However, the room-temperature strengths are much lower (=40 to 50%) than the room-temperature strength of hot–pressed Si₃N₄ (NC-132). The strength loss in Si₃N₄ and SiAlON materials at high temperature was attributed to a viscous grain-boundary phase retained during cooling from the sintering temperature. The oxidation resistance of sintered a-SiC was the best of any materials tested.     

Optoelectronic properties of nitrogenated amorphous carbon films synthesized by microwave surface wave plasma chemical vapor deposition system

The n-type nitrogen doped amorphous carbon (a-C:N) thin films have been grown by microwave (MW) surface wave plasma (SWP) chemical vapor deposition (CVD) system on silicon, quartz and ITO substrates at different nitrogen flow rates (1 to 4 sccm). The effects of nitrogen doping on chemical, optical, structural and electrical properties were studied through X-ray photoelectron spectroscopy, Nanopics 2100/NPX200 surface profiler, UV/VIS/NIR spectroscopy, Raman spectroscopy and solar simulator measurements. Argon, acetylene and nitrogen are used as plasma sources. Optical band gap decreased and nitrogen atomic concentration (%) increased with increasing nitrogen flow rate as a dopant. The a-C:N/p-Si based device exhibits photovoltaic behavior under illumination (AM 1.5, 100 mW/cm²), with a maximum open-circuit voltage (V_oc), short-circuit current (J_sc) and fill factor of 4.2 mV, 7.4 μA/cm² and 0.25 respectively.     

Refined memorization for vertex cover

Memorization is a technique which allows to speed up exponential recursive algorithms at the cost of an exponential space complexity. This technique already leads to the currently fastest algorithm for fixed-parameter vertex cover, whose time complexity is O(k^1.2832k^1.5+kn), where n is the number of nodes and k is the size of the vertex cover. Via a refined use of memorization, we obtain an O(k^1.2759k^1.5+kn) algorithm for the same problem. We moreover show how to further reduce the complexity to O(k^1.2745k⁴+kn).     

Scalable and Robust ANN Based Cooperative Spectrum Sensing for Cognitive Radio Networks

Cognitive radio network (CRN) supports dynamic spectrum access addressing spectrum scarcity issue experienced by today’s wireless communication network. Sensing is an important task and cooperative spectrum sensing is used for improving detection performance of spectrum. The sensing information from individual secondary users is sent to fusion center to infer a common global decision regarding primary user’s presence. Various fusion schemes for decision making are proposed in the literature but they lack scalability and robustness. We have introduced artificial neural network (ANN) at fusion center thereby achieving significant improvement in detection performance and reduction in false alarm rate as compared to conventional schemes. The proposed ANN scheme is found capable to deal with scalability of CRN with consistent performance. Further, SNR of individual Secondary user is taken into consideration in decision making at fusion center. Moreover the proposed scheme is tested against security attack (malicious users) and inadvertent errors occurring at SUs are found to be robust.     

Cluster-Head Restricted Energy Efficient Protocol (CREEP) for Routing in Heterogeneous Wireless Sensor Networks

A magnanimous number of collaborative sensor nodes make up a Wireless Sensor Network (WSN). These sensor nodes are outfitted with low-cost and low-power sensors. The routing protocols are responsible for ensuring communications while considering the energy constraints of the system. Achieving a higher network lifetime is the need of the hour in WSNs. Currently, many network layer protocols are considering a heterogeneous WSN, wherein a certain number of the sensors are rendered higher energy as compared to the rest of the nodes. In this paper, we have critically analysed the various stationary heterogeneous clustering algorithms and assessed their lifetime and throughput performance in mobile node settings also. Although many newer variants of Distributed Energy-Efficiency Clustering (DEEC) scheme execute proficiently in terms of energy efficiency, they suffer from high system complexity due to computation and selection of large number of Cluster Heads (CHs). A protocol in form of Cluster-head Restricted Energy Efficient Protocol (CREEP) has been proposed to overcome this limitation and to further improve the network lifetime by modifying the CH selection thresholds in a two-level heterogeneous WSN. Simulation results establish that proposed solution ameliorates in terms of network lifetime as compared to others in stationary as well as mobile WSN scenarios.