support vector machine (svm)

This study detects the activity of disabled people living in smart homes based on monitoring their behavior and desires to achieve the final model and store it in the database of the building management system for decision making and automatic operation of equipment used to meet the needs of residents. First, a suitable set of sensors are installed to record and process data correctly in different parts of the building and some to control vital signs depending on the type of disability are installed on the body. . First: selecting sensors and implementing recording techniques, and focusing on the major activities of people with disabilities in smart homes, second: classifying and processing the data collected using the support vector machine (svm) algorithm; svm was performed using the optimal data congestion algorithm (pso) and finally the final pattern for storage in the building management system database.

Prediction of flow discharge coefficient, Cd, for a sluice gate under free and submerged flow conditions is one of the essential issues in hydraulics. In recent years, various semi-empirical equations have been developed in order to predict Cd for a sluice gate that application of those formulas under submerged flow conditions suffered from large errors. The aim of the present research is to use Gaussian Process Regression (GPR) and Support Vector Machine (SVM) used in soft computing techniques, so that estimating Cd in submerged flow conditions and comparing the results with quasi-experimental methods are of interest, herein. For this purpose, an experimental dataset comprised of 122 data points were used to feed the methods utilized. Different combinations of dimensionless parameters were then prepared and the performance of the afore mentioned methods were assessed. The results showed that SVM with input parameters of 𝑦𝑡⁄𝑤, 𝑦0⁄𝑤, 1/𝐹𝑟2 and S by the values of Root Mean Square Error (RMSE=0.017), correlation coefficient (R=0.97) and Nash-Sutcliffe Equivalent (NSE=0.95) had a better performance than GPR and other semi-empirical approaches, indeed.

Today easy data access through the network has made it possible to steal them. Therefore, the security of computer systems has become increasingly important. Intrusion Detection Systems . as the last line of computer defense, can play an important role in attack resistance and their efficiencies has direct impact on network security. The Intrusion Detection Systems must extract the necessary strategies based on the connections and use them to detect new connections. Support Vector Machine is a Machine Learning method that it is popular to extract intrusion strategies in past decade. Although simplification of SVM returned it to popular method but it has constraints such as senility to kernel selection and it has not any optimization mechanism to determine the best of them. We model it as using of several kernels simultaneously and different weighting to them and dynamic SVM parameters. Due to the high complexity of this problem, conventional optimization methods are not able to solve it. Therefore, we propose a Distributed Self Adaptive Genetic Algorithm with Migration. On the other hand, due to the high volume of data in such issues, Autoencoder has been used to reduce data. The proposed approach is a hybrid method based on Autoencoder and improved Support Vector Machine with Distributed Self Adaptive Genetic Algorithm with Migration that it is evaluated by its execution on data set. The experimental results have demonstrated that the proposed system exhibits a high performance for attack detection based on precision and recall and it low time for intrusion.

In future smart grids, it's imperative to know the price of electricity market to guide the behavior of consumers and suppliers. This paper presents a hybrid approach for mid-term electricity price forecasting based on support vector machine and neural networks. In this method, at first, the price upper bound is considered. Then, the training set is divided into two parts including normal price and price spikes. Feature extraction applies on input data sets using stacked auto-encoders and a prediction model trained using each training set. Support Vector Machine (SVM) models with different kernel functions and a two layered feed-forward neural network were trained and tested with the proposed method. Simulation results using the proposed method show that this method has a significant effect on the speed of model training and improves forecasting accuracy.