A COVID-19 Detection Algorithm Using Deep Features and Discrete Social Learning Particle Swarm Optimization for Edge Computing Devices

COVID-19 has been spread around the world and has caused a huge number of deaths. Early detection of this disease is the most efficient way to prevent its rapid spread. Due to the development of internet technology and edge intelligence, developing an early detection system for COVID-19 in the medical environment of the Internet of Things (IoT) can effectively alleviate the spread of the disease. In this paper, a detection algorithm is developed, which can detect COVID-19 effectively by utilizing the features from Chest X-ray (CXR) images. First, a pre-trained model (ResNet18) is adopted for feature extraction. Then, a discrete social learning particle swarm optimization algorithm (DSLPSO) is proposed for feature selection. By filtering redundant and irrelevant features, the dimensionality of the feature vector is reduced. Finally, the images are classified by a Support Vector Machine (SVM) for COVID-19 detection. Experimental results show that the proposed algorithm can achieve competitive performance with fewer features, which is suitable for edge computing devices with lower computation power.

Amateur radio sensing technique using a combination of energy detection and waveform classification

<p>A critical problem in spectrum sensing is to create a detection algorithm and test statistics. The existing approaches employ the energy level of each channel of interest. However, this feature cannot accurately characterize the actual application of public amateur radio. The transmitted signal is not continuous and may consist only of a carrier frequency without information. This paper proposes a novel energy detection and waveform feature classification (EDWC) algorithm to detect speech signals in public frequency bands based on energy detection and supervised machine learning. The energy level, descriptive statistics, and spectral measurements of radio channels are treated as feature vectors and classifiers to determine whether the signal is speech or noise. The algorithm is validated using actual frequency modulation (FM) broadcasting and public amateur signals. The proposed EDWC algorithm's performance is evaluated in terms of training duration, classification time, and receiver operating characteristic. The simulation and experimental outcomes show that the EDWC can distinguish and classify waveform characteristics for spectrum sensing purposes, particularly for the public amateur use case. The novel technical results can detect and classify public radio frequency signals as voice signals for speech communication or just noise, which is essential and can be applied in security aspects.</p>

User Profiling Based on Nonlinguistic Audio Data

User profiling refers to inferring people’s attributes of interest ( AoIs ) like gender and occupation, which enables various applications ranging from personalized services to collective analyses. Massive nonlinguistic audio data brings a novel opportunity for user profiling due to the prevalence of studying spontaneous face-to-face communication. Nonlinguistic audio is coarse-grained audio data without linguistic content. It is collected due to privacy concerns in private situations like doctor-patient dialogues. The opportunity facilitates optimized organizational management and personalized healthcare, especially for chronic diseases. In this article, we are the first to build a user profiling system to infer gender and personality based on nonlinguistic audio. Instead of linguistic or acoustic features that are unable to extract, we focus on conversational features that could reflect AoIs. We firstly develop an adaptive voice activity detection algorithm that could address individual differences in voice and false-positive voice activities caused by people nearby. Secondly, we propose a gender-assisted multi-task learning method to combat dynamics in human behavior by integrating gender differences and the correlation of personality traits. According to the experimental evaluation of 100 people in 273 meetings, we achieved 0.759 and 0.652 in F1-score for gender identification and personality recognition, respectively.

Algorithm for detecting attacks on Web applications based on machine learning methods and attributes queries

Abstract—Almost developed applications tend to become as accessible as possible to the user on the Internet. Different applications often store their data in cyberspace for more effective work and entertainment, such as Google Docs, emails, cloud storage, maps, weather, news,... Attacks on Web resources most often occur at the application level, in the form of HTTP/HTTPS-requests to the site, where traditional firewalls have limited capabilities for analysis and detection attacks. To protect Web resources from attacks at the application level, there are special tools - Web Application Firewall (WAF). This article presents an anomaly detection algorithm, and how it works in the open-source web application firewall ModSecurity, which uses machine learning methods with 8 suggested features to detect attacks on web applications. Tóm tắt—Hầu hết các ứng dụng được phát triển có xu hướng trở nên dễ tiếp cận nhất có thể đối với người dùng qua Internet. Các ứng dụng khác nhau thường lưu trữ dữ liệu trên không gian mạng để làm việc và giải trí hiệu quả hơn, chẳng hạn như Google Docs, email, lưu trữ đám mây, bản đồ, thời tiết, tin tức,... Các cuộc tấn công vào tài nguyên Web thường xảy ra nhất ở tầng ứng dụng, dưới dạng các yêu cầu HTTP/HTTPS đến trang web, nơi tường lửa truyền thống có khả năng hạn chế trong việc phân tích và phát hiện các cuộc tấn công. Để bảo vệ tài nguyên Web khỏi các cuộc tấn công ở tầng ứng dụng, xuất hiện các công cụ đặc biệt - Tường lửa Ứng dụng Web (WAF). Bài viết này trình bày thuật toán phát hiện bất thường và cách thức hoạt động của tường lửa ứng dụng web mã nguồn mở ModSecurity khi sử dụng phương pháp học máy với 8 đặc trưng được đề xuất để phát hiện các cuộc tấn công vào các ứng dụng web.

Artificial neural networks for voice activity detection Technology

Currently, the direction of voice biometrics is actively developing, which includes two related tasks of recognizing the speaker by voice: the verification task, which consists in determining the speaker's personality, and the identification task, which is responsible for checking the belonging of the phonogram to a particular speaker. An open question remains related to improving the quality of the verification identification algorithms in real conditions and reducing the probability of error. In this work study Voice activity detection algorithm is proposed, which is a modification of the algorithm based on pitch statistics; VAD is investigated as a component of a speaker recognition system by voice, and therefore the main purpose of its work is to improve the quality of the system as a whole. On the example of the proposed modification of the VAD algorithm and the energy-based VAD algorithm, the analysis of the influence of the choice on the quality of the speaker recognition system is carried out.

Fiducial visibility on planar images during motion-synchronized tomotherapy treatments

Objective: Synchrony® is a motion management system on the Radixact® that uses planar kV radiographs to locate the target during treatment. The purpose of this work is to quantify the visibility of fiducials on these radiographs. Approach: A custom acrylic slab was machined to hold 8 gold fiducials of various lengths, diameters, and orientations with respect to imaging axis. The slab was placed on the couch at the imaging isocenter and planar radiographs were acquired perpendicular to the custom slab with varying thicknesses of acrylic on each side. Fiducial signal to noise ratio (SNR) and detected fiducial position error in millimeters were quantified. Main Results: The minimum output protocol (100 kVp, 0.8 mAs) was sufficient to detect all fiducials on both Radixact configurations when the thickness of the phantom was 20 cm. However, no fiducials for any protocol were detected when the phantom was 50 cm thick. The algorithm accurately detected fiducials on the image when the SNR was larger than 4. The MV beam was observed to cause RFI artifacts on the kV images and to decrease SNR by an average of 10%. Significance: This work provides the first data on fiducial visibility on kV radiographs from Radixact Synchrony treatments. The Synchrony fiducial detection algorithm was determined to be very accurate when sufficient SNR is achieved. However, a higher output protocol may need to be added for use with larger patients. This work provided groundwork for investigating visibility of fiducial-free solid targets in future studies and provided a direct comparison of fiducial visibility on the two Radixact configurations, which will allow for intercomparison of results between configurations.