China
Africa
Explore chapters and articles related to this topic
Machine Learning in the Steel Industry
Published in Monika Mangla, Subhash K. Shinde, Vaishali Mehta, Nonita Sharma, Sachi Nandan Mohanty, Handbook of Research on Machine Learning, 2022
The second method is the SVM based method. In this method, the SVM is first trained with signals of cobble and no-cobble data. The trained two-class classifier is used to predict the possibility of cobble generation in the mill. SVM was selected due to its simplicity, robustness to the outliers, and efficiency in binary classification. The kernel function and associated parameters are calibrated empirically to obtain the best training and testing performances. The available datasets were randomly divided into two sets-training and testing: 70% of datasets for training and the remaining 30% of datasets for testing before training and testing. A total number of 255 features are generated from the raw dataset for the training of SVM. However, all these 255 features may not be relevant for the classification of the existence of cobble into two classes-Cobble and No-Cobble. Therefore, 113 features out of the available 255 features were selected based on mutual information. A threshold for mutual selection is selected empirically, the features carrying higher mutual information with respect to the class labels are selected for this purpose. After training of the SVM classifier, it was used to predict cobble generation using online real-time signals. Both the statistical and SVM classifiers were superimposed to predict cobble with high reliability. A partial screenshot of the software screen is shown in Figure 12.11.
Literature review and proposed framework
Published in Juan Carlos Chacon-Hurtado, Optimisation of Dynamic Heterogeneous Rainfall Sensor Networks in the Context of Citizen Observatories, 2019
Mutual information is a measurement of the amount of information that a variable contains about another. This is measured as the relative entropy between the joint distribution and the product distribution (Cover and Thomas 2006). In the simplest expression (two variables), the mutual information can be defined as: I(X1,X2)=H(X1)+H(X2)−H(X1,X2)
Entropy based multi-criteria evaluation for rainfall monitoring networks under seasonal effect
Published in Chongfu Huang, Zoe Nivolianitou, Risk Analysis Based on Data and Crisis Response Beyond Knowledge, 2019
Heshu Li, Dong Wang*, Yuankun Wang
Mutual information is defined to estimate the shared information between two variables X and Y, and it can be interpreted as the reduction in the uncertainty of X given the knowledge of Y: IX,Y=∑i=1m∑j=1npxi,yjlog2pxi,yjpxipyj
LGBM-based modeling scenarios to compressive strength of recycled aggregate concrete with SHAP analysis
Published in Mechanics of Advanced Materials and Structures, 2023
Bin Xi, Enming Li, Yewuhalashet Fissha, Jian Zhou, Pablo Segarra
To obtain a nonlinear relationship between the variables, mutual information regression coefficients were used. The method quantifies the shared information between input features and output variables [27]. Specifically, it measures how much information about one variable can be gained by observing the other variable. Mutual information regression can capture non-linear relationships, handle both categorical and continuous features, and is robust to noise. The mutual information coefficient can take on values ranging from 0 to infinity, where higher values indicate a greater amount of shared information between the variables. In Figure 2(b), it can be observed that the variables RAC and FAC exhibit the highest values of interaction information on the variable CS. This suggests that these two variables have a stronger relationship with CS compared to other variables in the dataset. This finding is significant because it cannot be discerned from the results obtained through the Pearson correlation coefficient method.
Preventing Reverse Engineering of Critical Industrial Data with DIOD
Published in Nuclear Technology, 2023
Arvind Sundaram, Hany S. Abdel-Khalik, Mohammad G. Abdo
A formal notion to describe the information between two variables and is the mutual information that exists for many types of variables including categorical data, real numbers, continuous variables, discrete variables, etc. Developed using information theory, the mutual information may be defined as the reduction in the entropy of one variable given the other. The mutual information is also invariant to invertible smooth nonlinear transformations and as described in Eq. (6):
Hybrid three-dimensional modelling for reservoir fluorescent dissolved organic matter risk assessment
Published in Inland Waters, 2022
Xinchen Wang, Hong Zhang, Edoardo Bertone, Rodney A. Stewart, Sara P. Hughes
The sensitivity analysis was conducted for both naïve BNs (Fig. 7), and Shannon’s mutual information (Shannon and Warren 1949), one of the most commonly used metrics to rank information sources, was calculated. In simple terms, it represents the total uncertainty-reducing potential of the predictor in relation to the targeted variable (Pearl 1988; i.e., the larger the mutual information, the larger the uncertainty-reducing potential of that predictor and thus the more sensitive the target variable is to that predictor). Mutual information is a nonnegative number, equal to zero when the target and predictor are mutually independent (Gallager 1968). The peak fDOM value is highly sensitive to rainfall/inflow-related variables, especially the rainfall amount and the inflow from Tingalpa Creek. However, the time between the peak rainfall and the peak in fDOM at the dam wall is most sensitive to reservoir volume, followed by residence time and wind speed, affecting mixing process. Rainfall and inflow variables are less important for predicting fDOM peak value. Wind directions and duration are overall less sensitive variables for both models.