scholarly journals Efficiently Classifying Lung Sounds through Depthwise Separable CNN Models with Fused STFT and MFCC Features

Diagnostics ◽  
2021 ◽  
Vol 11 (4) ◽  
pp. 732
Author(s):  
Shing-Yun Jung ◽  
Chia-Hung Liao ◽  
Yu-Sheng Wu ◽  
Shyan-Ming Yuan ◽  
Chuen-Tsai Sun
Keyword(s):  
Neural Network ◽  
Lung Diseases ◽  
Medical Professionals ◽  
Feature Engineering ◽  
Model Design ◽  
Lung Sounds ◽  
Engineering Process ◽  
The World ◽  
Lung Auscultation ◽  
Short Time

Lung sounds remain vital in clinical diagnosis as they reveal associations with pulmonary pathologies. With COVID-19 spreading across the world, it has become more pressing for medical professionals to better leverage artificial intelligence for faster and more accurate lung auscultation. This research aims to propose a feature engineering process that extracts the dedicated features for the depthwise separable convolution neural network (DS-CNN) to classify lung sounds accurately and efficiently. We extracted a total of three features for the shrunk DS-CNN model: the short-time Fourier-transformed (STFT) feature, the Mel-frequency cepstrum coefficient (MFCC) feature, and the fused features of these two. We observed that while DS-CNN models trained on either the STFT or the MFCC feature achieved an accuracy of 82.27% and 73.02%, respectively, fusing both features led to a higher accuracy of 85.74%. In addition, our method achieved 16 times higher inference speed on an edge device and only 0.45% less accuracy than RespireNet. This finding indicates that the fusion of the STFT and MFCC features and DS-CNN would be a model design for lightweight edge devices to achieve accurate AI-aided detection of lung diseases.

scholarly journals p1-FP: Extraction, Classification, and Prediction of Website Fingerprints with Deep Learning

2019 ◽  
Vol 2019 (3) ◽  
pp. 191-209 ◽  
Author(s):  
Se Eun Oh ◽  
Saikrishna Sunkam ◽  
Nicholas Hopper
Keyword(s):  
Neural Network ◽  
Machine Learning ◽  
Deep Learning ◽  
Deep Neural Network ◽  
State Of The Art ◽  
Machine Learning Techniques ◽  
Feature Engineering ◽  
Engineering Process ◽  
Learning Techniques ◽  
Wide Range

Abstract Recent advances in Deep Neural Network (DNN) architectures have received a great deal of attention due to their ability to outperform state-of-the-art machine learning techniques across a wide range of application, as well as automating the feature engineering process. In this paper, we broadly study the applicability of deep learning to website fingerprinting. First, we show that unsupervised DNNs can generate lowdimensional informative features that improve the performance of state-of-the-art website fingerprinting attacks. Second, when used as classifiers, we show that they can exceed performance of existing attacks across a range of application scenarios, including fingerprinting Tor website traces, fingerprinting search engine queries over Tor, defeating fingerprinting defenses, and fingerprinting TLS-encrypted websites. Finally, we investigate which site-level features of a website influence its fingerprintability by DNNs.

scholarly journals Preliminary Detection of Lung Diseases in Pediatric Population using Soft Computing

2019 ◽  
Vol 8 (3) ◽  
pp. 2731-2735
Keyword(s):  
Neural Network ◽  
Artificial Neural Network ◽  
Lung Diseases ◽  
Human Subjects ◽  
Pediatric Population ◽  
Lung Sounds ◽  
Electronic Stethoscope ◽  
Temporal Features ◽  
Lung Abnormalities ◽  
Hidden Layer

The investigations from recent studies clearly show the potential of lung sounds in detection of lung abnormalities in human subjects. This paper aims to analyze lung sounds acquired using special electronic stethoscope for detection adventitious sounds arising out of pathological lungs due to various disease like brochities especially in pediatric population. For acquisition and recording of lung sounds, 3M Littmann 3200 model is utilized. After verifying fidelity of electronic stethoscope, the analysis of lung sounds was carried out by various spectral and temporal features. The features extracted were fed to artificial neural network for classification. Various combinations of ANN with different topologies were experimented. The overall accuracy of obtained with one hidden layer GFF is 94.95%.

Development of Dermatological Skin Disease Detection and Classification based on Wavelet and ANN

2018 ◽  
Vol 8 (5) ◽  
pp. 10
Author(s):  
Revati Kadu ◽  
U. A. Belorkar
Keyword(s):  
Neural Network ◽  
Artificial Neural Network ◽  
Skin Disease ◽  
Skin Diseases ◽  
Screening Method ◽  
Phase Detection ◽  
Statistical Measures ◽  
The World ◽  
Artificial Neural ◽  
Wavelet Techniques

One of the most common and augmenting health problems in the world are related to skin. The most  unpredictable and one of the most difficult entities to automatically detect and evaluate is the human skin disease because of complexities of texture, tone, presence of hair and other distinctive features. Many cases of skin diseases in the world have triggered a need to develop an effective automated screening method for detection and diagnosis of the area of disease. Therefore the objective of this work is to develop a new technique for automated detection and analysis of the skin disease images based on color and texture information for skin disease screening. In this paper, system is proposed which detects the skin diseases using Wavelet Techniques and Artificial Neural Network. This paper presents a wavelet-based texture analysis method for classification of five types of skin diseases. The method applies tree-structured wavelet transform on different color channels of red, green and blue dermoscopy images, and employs various statistical measures and ratios on wavelet coefficients. In all 99 unique features are extracted from the image. By using Artificial Neural Network, the system successfully detects different types of dermatological skin diseases. It consists of mainly three phases image processing, training phase, detection  and classification phase.

PERCEPTION AND INTERPRETATION OF CREATIVITY OF ALISHER NAVOI IN FRENCH ORIENTAL STUDIES

2020 ◽  
Vol 6 (3) ◽  
pp. 40-50
Author(s):  
Mukhammadjon Holbekov ◽  
Keyword(s):  
The World ◽  
Oriental Studies ◽  
Short Time ◽  
The Universe ◽  
Precious Stones

The great Uzbek poet Alisher Navoi(1441-1501), during his lifetime, was widely known not only in his homeland, but also far beyond its borders. A contemporary and biographer of Navoi, the famous historian Hondemir, of course, not without some hyperbole, wrote: "He (Navoi -M.Kh.) in a short time took the cane of primacy from his peers; the fame of his talents spread to all ends of the world, and the stories of the firmness of his noble mind from mouth to mouth were innumerable.The pearls of his poetry adorned the leaves of the Book of Fates, the precious stones of his poetry filled the shells of the universe with pearls of beauty

Low-Cost Hardware Implementation of Human Blood Identification Device Using Feed-Forward Artificial Neural Network on FPGA

2018 ◽  
Author(s):  
Rizki Eka Putri ◽  
Denny Darlis
Keyword(s):  
Neural Network ◽  
Artificial Neural Network ◽  
Low Cost ◽  
Image Resolution ◽  
Blood Type ◽  
Feed Forward ◽  
Testing Facility ◽  
System Input ◽  
Artificial Neural ◽  
Short Time

This article was under review for ICELTICS 2018 -- In the medical world there is still service dissatisfaction caused by lack of blood type testing facility. If the number of tested blood arise, a lot of problems will occur so that electronic devices are needed to determine the blood type accurately and in short time. In this research we implemented an Artificial Neural Network on Xilinx Spartan 3S1000 Field Programable Gate Array using XSA-3S Board to identify the blood type. This research uses blood sample image as system input. VHSIC Hardware Discription Language is the language to describe the algorithm. The algorithm used is feed-forward propagation of backpropagation neural network. There are 3 layers used in design, they are input, hidden1, and output. At hidden1layer has two neurons. In this study the accuracy of detection obtained are 92%, 92%, 92%, 90% and 86% for 32x32, 48x48, 64x64, 80x80, and 96x96 pixel blood image resolution, respectively.

scholarly journals Efficacy of compressed sodium chloride (CSC) against E. coli and Candida auris in minutes and methods improvement for testing

2021 ◽  
Vol 11 (1) ◽  
Author(s):  
Lan N. Truong ◽  
Brayden D. Whitlock
Keyword(s):  
Sodium Chloride ◽  
Improved Method ◽  
Candida Auris ◽  
Antimicrobial Surfaces ◽  
E Coli ◽  
Route Of Transmission ◽  
The World ◽  
Short Time ◽  
First Time ◽  
Time Point

AbstractControlling infections has become one of the biggest problems in the world, whether measured in lives lost or money spent. This is worsening as pathogens continue becoming resistant to therapeutics. Antimicrobial surfaces are one strategy being investigated in an attempt to decrease the spread of infections through the most common route of transmission: surfaces, including hands. Regulators have chosen two hours as the time point at which efficacy should be measured. The objectives of this study were to characterize the new antimicrobial surface compressed sodium chloride (CSC) so that its action may be understood at timepoints more relevant to real-time infection control, under two minutes; to develop a sensitive method to test efficacy at short time points; and to investigate antifungal properties for the first time. E. coli and Candida auris are added to surfaces, and the surfaces are monitored by contact plate, or by washing into collection vats. An improved method of testing antimicrobial efficacy is reported. Antimicrobial CSC achieves at least 99.9% reduction of E. coli in the first two minutes of contact, and at least 99% reduction of C. auris in one minute.

Neural Network-Based Model Design for Short-Term Load Forecast in Distribution Systems

2016 ◽  
Vol 31 (1) ◽  
pp. 72-81 ◽  
Author(s):  
Ni Ding ◽  
Clementine Benoit ◽  
Guillaume Foggia ◽  
Yvon Besanger ◽  
Frederic Wurtz
Keyword(s):  
Neural Network ◽  
Distribution Systems ◽  
Model Design ◽  
Short Term ◽  
Load Forecast

scholarly journals Validating Deep Neural Networks for Online Decoding of Motor Imagery Movements from EEG Signals

Sensors ◽  
2019 ◽  
Vol 19 (1) ◽  
pp. 210 ◽  
Author(s):  
Zied Tayeb ◽  
Juri Fedjaev ◽  
Nejla Ghaboosi ◽  
Christoph Richter ◽  
Lukas Everding ◽  
...  
Keyword(s):  
Neural Network ◽  
Machine Learning ◽  
Deep Learning ◽  
Convolutional Neural Network ◽  
Motor Imagery ◽  
Classification Performance ◽  
Feature Engineering ◽  
Learning Models ◽  
Eeg Signals ◽  
Learning Methods

Non-invasive, electroencephalography (EEG)-based brain-computer interfaces (BCIs) on motor imagery movements translate the subject’s motor intention into control signals through classifying the EEG patterns caused by different imagination tasks, e.g., hand movements. This type of BCI has been widely studied and used as an alternative mode of communication and environmental control for disabled patients, such as those suffering from a brainstem stroke or a spinal cord injury (SCI). Notwithstanding the success of traditional machine learning methods in classifying EEG signals, these methods still rely on hand-crafted features. The extraction of such features is a difficult task due to the high non-stationarity of EEG signals, which is a major cause by the stagnating progress in classification performance. Remarkable advances in deep learning methods allow end-to-end learning without any feature engineering, which could benefit BCI motor imagery applications. We developed three deep learning models: (1) A long short-term memory (LSTM); (2) a spectrogram-based convolutional neural network model (CNN); and (3) a recurrent convolutional neural network (RCNN), for decoding motor imagery movements directly from raw EEG signals without (any manual) feature engineering. Results were evaluated on our own publicly available, EEG data collected from 20 subjects and on an existing dataset known as 2b EEG dataset from “BCI Competition IV”. Overall, better classification performance was achieved with deep learning models compared to state-of-the art machine learning techniques, which could chart a route ahead for developing new robust techniques for EEG signal decoding. We underpin this point by demonstrating the successful real-time control of a robotic arm using our CNN based BCI.

Grinding Acoustic Emission Classification in Terms of Mechanical Behaviours

2007 ◽  
Vol 329 ◽  
pp. 15-20 ◽  
Author(s):  
Xun Chen ◽  
James Griffin
Keyword(s):  
Neural Network ◽  
Acoustic Emission ◽  
Material Removal ◽  
Grinding Process ◽  
Short Time Fourier Transform ◽  
Fundamental Investigation ◽  
Mechanical Behaviours ◽  
Wavelets Transform ◽  
Short Time ◽  
Ae Signals

The material removal in grinding involves rubbing, ploughing and cutting. For grinding process monitoring, it is important to identify the effects of these different phenomena experienced during grinding. A fundamental investigation has been made with single grit cutting tests. Acoustic Emission (AE) signals would give the information relating to the groove profile in terms of material removal and deformation. A combination of filters, Short-Time Fourier Transform (STFT), Wavelets Transform (WT), statistical windowing of the WT with the kurtosis, variance, skew, mean and time constant measurements provided the principle components for classifying the different grinding phenomena. Identification of different grinding phenomena was achieved from the principle components being trained and tested against a Neural Network (NN) representation.

Optimization of the Robots Fourier Spectrum by Using the Assisted Research, Neural Network, Smart Damper and LabVIEW Instrumentation

2012 ◽  
Vol 245 ◽  
pp. 24-32 ◽  
Author(s):  
Adrian Olaru ◽  
Serban Olaru ◽  
Aurel Oprean
Keyword(s):  
Neural Network ◽  
Transfer Function ◽  
Spectral Density ◽  
Vibration Analysis ◽  
Manufacturing Systems ◽  
Fourier Spectrum ◽  
Industrial Robots ◽  
Vibration Behavior ◽  
Power Spectral ◽  
The World

The most important things in the dynamic research of industrial robots are the vibration behavior, the transfer function and the vibration power spectral density between some of the robot joints and components. In the world this research is made without the assisted research. In each of the study cases in this paper was used the proper virtual Fourier analyzer and was presented one new method of the assisted vibration analysis. With this research it is possible the optimal choosing the base modulus type to avoid the frequencies from the robot spectrum. In the manufacturing systems, the most important facts are the vibration behavior of the robot, the compatibility with some other components of the system. All the VI where achieved in the LabVIEW soft 8.2 version, from National Instruments, USA. This method and the created virtual LabVIEW instrumentation are generally and they are possible to apply in many other dynamic behavior research.

Sign in / Sign up

Export Citation Format

Share Document