China 
Africa 
Explore chapters and articles related to this topic
Parameter Estimation
Published in S. Sitharama Iyengar, Richard R. Brooks, Distributed Sensor Networks, 2016
The matrices U^, ∑^, and V^T are provided to the individual platforms. When a target passes near the platform, the time series data is process and matched against the reduced-dimensional vector database as described above and shown in the online SIF algorithm, Figure 7.8.
Content-Based Feature Extraction: Image Transforms
Published in Rik Das, Content-Based Image Classification, 2020
The following steps are used for classification with feature extraction using partial energy coefficients:Red, Green and Blue color components are extracted from a given image.Image Transform is applied on each of the components to extract feature vectors.The extracted feature vectors from each of the components are stored as a complete set of feature vectors.Further, partial coefficients from the entire feature vector set are extracted to form the feature vector database.The feature vector database, with 100% transformed coefficients and partial coefficients ranging from 50% of the complete set of feature vectors till 0.06% of the complete set of feature vectors, is constructed as shown in Fig. 5.1.The feature vectors of the query image for the whole set of feature vectors and for partial coefficient of feature vectors are compared with the database images for classification results.Classifications are done with the entire training set for the query images and are compared for the highest classification result to find the best percentage of partial coefficient for feature extraction technique.Consider the best percentage thus inferred as the feature set extracted by applying image transform.
Medical and Mathematical Background
Published in Arwa Ahmed Gasm Elseid, Alnazier Osman Mohammed Hamza, Computer-Aided Glaucoma Diagnosis System, 2020
Arwa Ahmed Gasm Elseid, Alnazier Osman Mohammed Hamza
This system analyzes the feature vector using multiple sets of rules that are designed to test specific conditions in the feature vector database to set off an action. The rules consist of two parts: condition premises and actions, which are generated based on expert knowledge to decide the action when the conditions are satisfied, where the action taken will be a part of the rule that could change or a labeling of the feature vector based on the analysis.
Separation of Machine-Printed and Handwritten Texts in Noisy Documents using Wavelet Transform
Published in IETE Technical Review, 2019
In this step, two texture features described in Sections 3.1 and 3.2 are computed from the training databases. The resultant feature vectors are stored in feature vector database. Feature vector is written as where V and D are BVLCC and BDIP moments, respectively. Initially, classifier is trained using these features and then classification is performed for the query is whether for printed text, handwritten text, or noise. For classification, SVM [47,48] is used shown in Figure 5. The concept of SVM has strong connection with the statistical learning theory. SVM is a supervised binary classifier whose main work is to find the optimum hyperplane separating data points of different classes in high dimensional space.
Identifying disaster related social media for rapid response: a visual-textual fused CNN architecture
Published in International Journal of Digital Earth, 2020
Xiao Huang, Zhenlong Li, Cuizhen Wang, Huan Ning
Given the fact that textual patterns differ a lot in short-text posts in social media compared to formal sources including news and formal articles, it is necessary to train word vectors specifically for social media posts. To acquire word vectors, the technique used in this study is Word2Vec, a shallow neural network with single hidden layer, but proved to be powerful in providing 300-dimention vectors representing the word characteristics (Mikolov et al. 2013). A word vector database is then built to enable the formation of sentence matrix, which is the input for a CNN architecture designed based on the work by Kim (2014) (Figure 3).