Convolution is considered most significant layer in deep learning because it can extract best features of data through the network but it may result in huge volume of data. This problem can be solved by using pooling. In this paper, A novel pooling method is proposed by using discrete Fourier transform (DFT), this method is used DFT technique to transform the data from spatial domain into frequency domain to preserve the most important information from the details coefficients, where the details information of the image is less significant, therefore it can be discarded to down sample the size of dimensions. Its effect will be great with advantage of reducing the eliminated details information as compared with other standard methods. After applying DFT, the most significant coefficients, which represent most important features are cropped while less important details will be discarded then the data are reconstructed by applying inverse DF, therefore the high quality of features are extracted, which solve the problem of losing significant information during the pooling layer. Different methods are proposed based on the scenario of using DFT. The proposed methods are tested by extracting pooled image then the original images were retrieved using only the pooled images. Then the retrieved images are compared with original images by using different measures such as SNR, correlation and SSIM. Then the proposed layers used for image classification for two different datasets. The results proved that the proposed methods outperformed standard methods, thus it can be used for deep learning application.

No abstract

IOT information is always subjected to attacks, because component of the IOT system always unsupervised for most of time, also due to simplicity of wireless communication media, so there is high chance for attack, lastly, IOT is constraint device in terms of energy and computation complexity. So, different research and study are proposed to provide cryptographic algorithm. In this paper, a new image encryption is proposed based on anew chaotic map used to generate the binary key. The proposed map is three dimensional map, which is more sensitive to initial condition, each dimension of the 3-D chaotic map is depended on the others dimension, which may increase the randomness of the behavior trajectory for the next values and this gives the algorithm the ability to resist any attacks. At first, 3-D chaotic map is proposed, which is very sensitive for initial condition, the three dimension is depended on each other, which make the system more randomness, then the produced sequences is converted on binary key by using mod operation. The original image is scrambled based on mod operation to exchange the row and interleaving them, the same operations are repeated for column of the image. Later, the image is divided into blocks of size (8*8) and scrambled by using negative diagonal scan, the final pixels are converted into binary sequences, which are XORed with the generated key to produce the encrypted image. The experiment is performed on different images with different properties and tested with different metrics such as entropy, correlation, key sensitivity, number of pixel change rate (NPCR) and histogram of the original and encrypted images. T results shows that the proposed encryption algorithm is more efficient and outperform other methods.

Real time monitoring with IOT is developed in the industry of health care , this can enable the doctors and specialist to diagnosis the patient status in quick, smart and efficient methods. Although, there is a lot of research and studies are designed methods for observing the ECG signal remotely, there are no proposed methods for classifying these signals with monitoring, and therefore , to design complete health care system, classification techniques should be used to classify the extracted signal. In this paper , We have proposed ECG monitoring and classification system. The proposed system is extracted  ECG signal based on AD8232 sensor  with the ardunino nodeMcu, analog to digital converter and its communication is used to convert the signal to more precision , then  the extracted signal is transmitted to cloud to be used at anywhere by using cloud, the signal is pre-processed to remove the noise and QRS complex is detected to determine the other characteristics of the signal such as heart rate, also to determine one cycle of ECG signal, later the signal is classified by using proposed  convolution neural network model  to detect the signal status. The extracted ECG signal is transmitted in real time to cloud (Ubidots cloud is used) through ESP8266 over to the cloud using WiFi based on MQTT publishing method. The experimental results are performed on different signals and the different stage of de-noising and QRS detection are applied and  different pooling layers are used in the proposed CNN model.  The results show that the proposed classification model is achieved  accuracy (94.94%) with  ( 94.56%), (94.56% ) and ( 5.06) for sensitivity, specificity and error rate (ERR)  respectively

Heart disease is the leading cause of death, the cardiovascular disease (CVD) is the major cause of the death world wide according to world health organization. Over 30% of global death was because CVD. However it is considered as controllable disease, so early and accurate diagnosis of heart disease is essential to administrating early and optimal treatment in order to increase long –term survival. Early detection can lead to reduce disease progression. In this paper, we propose a new deep neural network that can be used as classifier in heart disease prediction system, the data base is splitted into training and testing parts, the training data are prepressed by extracting its features in order to perform data augmentation, then the augmented data are training by the designed new model that can increase the accuracy of heart disease detection. from the experimental results, the proposed model provide significant improvement in the prediction of the disease in terms of accuracy, sensitivity and specificity as compared with other approaches