Neural Network
Recently Published Documents


(FIVE YEARS 87323)



2021 ◽  
Vol 21 (5) ◽  
pp. 49-57
Dongwook Kim ◽  
Cho-Rok Jang ◽  
Jung-Yun Cho ◽  
Moon-Yup Jang ◽  
Juil Song

Recently, the incidence of heat waves has increased due to climate change, and the resultant mortalities and socio-economic damage are also increasing in Korea. Hence, emphasis has been placed on research examining heatwaves and their effects. Predicting the probability of heatwave in advance is very important from the perspective of disaster risk management; however, related studies have been insufficient so far. Therefore, in this study, the probability of future heatwave onset was predicted using daily scaled past weather data for Seoul Metropolitan Government. For the analysis, models based on recurrent neural networks (RNN, LSTM, GRU) were used, which are suitable for analyzing time-series data. Upon evaluating the performance of the GRU model, which was selected as the optimized model, no overfitting problem was observed. The prediction accuracy of the model was high as it demonstrated a reproduction of 78% and 86% of actual heatwave days during the validation and test process, respectively. Therefore, this model can be used by each local government to coordinate an efficient response to heat waves.

2022 ◽  
Vol 16 (1) ◽  
pp. 0-0

Anomaly detection is a very important step in building a secure and trustworthy system. Manually it is daunting to analyze and detect failures and anomalies. In this paper, we proposed an approach that leverages the pattern matching capabilities of Convolution Neural Network (CNN) for anomaly detection in system logs. Features from log files are extracted using a windowing technique. Based on this feature, a one-dimensional image (1×n dimension) is generated where the pixel values of an image correlate with the features of the logs. On these images, the 1D Convolution operation is applied followed by max pooling. Followed by Convolution layers, a multi-layer feed-forward neural network is used as a classifier that learns to classify the logs as normal or abnormal from the representation created by the convolution layers. The model learns the variation in log pattern for normal and abnormal behavior. The proposed approach achieved improved accuracy compared to existing approaches for anomaly detection in Hadoop Distributed File System (HDFS) logs.

2022 ◽  
Vol 12 (1) ◽  
pp. 1-11
Sanat Kumar Sahu ◽  
Pratibha Verma

In this paper, Feature Selection Technique (FST) namely Particle Swarm Optimization (PSO) has been used. The filter based PSO is a search method with Correlation-based Feature Selection (CBFS) as a fitness function. The FST has two key goals of improving classification efficiency and reducing feature counts. Artificial Neural Network (ANN) Based Multilayer Perceptron Network (MLP) and Deep Learning (DL) have been considered the classification methods on 2 benchmark Autistic Spectrum Disorder (ASD) dataset. The experimental result was compared to the non-reduced features and reduced feature of ASD datasets. The reduced feature give up enhanced results in both classifiers MLP and DL. In addition, an experimental study on the exhibitions of these methodologies has been conducted. Finally, a new trend of PSO-MLP and PSO-DL based classification model is proposed.

2021 ◽  
Vol ahead-of-print (ahead-of-print) ◽  
Daoling Chen ◽  
Pengpeng Cheng

PurposeIn order to help companies better grasp the perceptual needs of consumers for patterns, so as to carry out more accurate product pattern development and recommendation, this research develops a product pattern design system based on computer-aided design.Design/methodology/approachFirst, use the Kansei engineering theory and method to obtain the user's perceptual image, and deconstruct and encode the pattern based on the morphological analysis method, then through the BP neural network to construct the mapping relationship between the user's perceptual image and the pattern design elements, and finally calculate and find the corresponding design code combination according to the design goal to guide the pattern design.FindingsTaking costume paper-cut patterns as an example, the feasibility of this system is verified, the design system can well reflect the user's perceptual image in the pattern design and improve the efficiency of pattern customization service.Originality/valueCompared with the traditional method that relies on the designer's personal experience to propose a design plan, this research provides scientific and intelligent design methods for product pattern design.

Aavani B

Abstract: Diabetic retinopathy is the leading cause of blindness in diabetic patients. Screening of diabetic retinopathy using fundus image is the most effective way. As the time increases this DR leads to permanent loss of vision. At present, Diabetic retinopathy is still being treated by hand by an ophthalmologist which is a time-consuming process. Computer aided and fully automatic diagnosis of DR plays an important role in now a day. Data-set containing a collection of fundus images of different severity scale is used to analyze the fundus image of DR patients. Here the deep neural network model is trained by using this fundus image and five-degree classification task is performed. We were able to produce an sensitivity of 90%. Keywords: Confusion matrix, Deep convolutional Neural Network, Diabetic Retinopathy, Fundus image, OCT

Ambeshwar Kumar ◽  
Ramachandran Manikandan ◽  
Utku Kose ◽  
Deepak Gupta ◽  
Suresh C. Satapathy

In Medicine Deep Learning has become an essential tool to achieve outstanding diagnosis on image data. However, one critical problem is that Deep Learning comes with complicated, black-box models so it is not possible to analyze their trust level directly. So, Explainable Artificial Intelligence (XAI) methods are used to build additional interfaces for explaining how the model has reached the outputs by moving from the input data. Of course, that's again another competitive problem to analyze if such methods are successful according to the human view. So, this paper comes with two important research efforts: (1) to build an explainable deep learning model targeting medical image analysis, and (2) to evaluate the trust level of this model via several evaluation works including human contribution. The target problem was selected as the brain tumor classification, which is a remarkable, competitive medical image-based problem for Deep Learning. In the study, MR-based pre-processed brain images were received by the Subtractive Spatial Lightweight Convolutional Neural Network (SSLW-CNN) model, which includes additional operators to reduce the complexity of classification. In order to ensure the explainable background, the model also included Class Activation Mapping (CAM). It is important to evaluate the trust level of a successful model. So, numerical success rates of the SSLW-CNN were evaluated based on the peak signal-to-noise ratio (PSNR), computational time, computational overhead, and brain tumor classification accuracy. The objective of the proposed SSLW-CNN model is to obtain faster and good tumor classification with lesser time. The results illustrate that the SSLW-CNN model provides better performance of PSNR which is enhanced by 8%, classification accuracy is improved by 33%, computation time is reduced by 19%, computation overhead is decreased by 23%, and classification time is minimized by 13%, as compared to state-of-the-art works. Because the model provided good numerical results, it was then evaluated in terms of XAI perspective by including doctor-model based evaluations such as feedback CAM visualizations, usability, expert surveys, comparisons of CAM with other XAI methods, and manual diagnosis comparison. The results show that the SSLW-CNN provides good performance on brain tumor diagnosis and ensures a trustworthy solution for the doctors.

2021 ◽  
Vol ahead-of-print (ahead-of-print) ◽  
Wenda Wei ◽  
Chengxia Liu ◽  
Jianing Wang

PurposeNowadays, most methods of illusion garment evaluation are based on the subjective evaluation of experienced practitioners, which consumes time and the results are too subjective to be accurate enough. It is necessary to explore a method that can quantify professional experience into objective indicators to evaluate the sensory comfort of the optical illusion skirt quickly and accurately. The purpose of this paper is to propose a method to objectively evaluate the sensory comfort of optical illusion skirt patterns by combining texture feature extraction and prediction model construction.Design/methodology/approachFirstly, 10 optical illusion sample skirts are produced, and 10 experimental images are collected for each sample skirt. Then a Likert five-level evaluation scale is designed to obtain the sensory comfort level of each skirt through the questionnaire survey. Synchronously, the coarseness, contrast, directionality, line-likeness, regularity and roughness of the sample image are calculated based on Tamura texture feature algorithm, and the mean, contrast and entropy are extracted of the image transformed by Gabor wavelet. Both are set as objective parameters. Two final indicators T1 and T2 are refined from the objective parameters previously obtained to construct the predictive model of the subjective comfort of the visual illusion skirt. The linear regression model and the MLP neural network model are constructed.FindingsResults show that the accuracy of the linear regression model is 92%, and prediction accuracy of the MLP neural network model is 97.9%. It is feasible to use Tamura texture features, Gabor wavelet transform and MLP neural network methods to objectively predict the sensory comfort of visual illusion skirt images.Originality/valueCompared with the existing uncertain and non-reproducible subjective evaluation of optical illusion clothing based on experienced experts. The main advantage of the authors' method is that this method can objectively obtain evaluation parameters, quickly and accurately obtain evaluation grades without repeated evaluation by experienced experts. It is a method of objectively quantifying the experience of experts.

2022 ◽  
Vol 32 (1) ◽  
pp. 87-99
Meenakshi Malik ◽  
Rainu Nandal ◽  
Surjeet Dalal ◽  
Vivek Jalglan ◽  
Dac-Nhuong Le

Sign in / Sign up

Export Citation Format

Share Document