Читать книгу Machine Learning Paradigm for Internet of Things Applications - Группа авторов - Страница 44

ML allows a computer to automatically learn and grow (without directly being programmed). Overall, ML algorithms can be classified as being (i) managed, (ii) unmonitored, and (iii) evolutionary computation. In reinforcement methods, a description is associated with each input value, while input values remain unlabeled in unsupervised learning. Learning algorithm implements a reinforcement-based mechanism in which the goal is to choose the set of environmental activities that optimize the overall benefit. There are some several types of machine learning algorithms, namely, SVM (Support Vector Machine). For both classification and regression queries [33], SVMs are valid, but they are widely used for the former. A binary SVM performs a binary division, generating a hyperplane such that it is possible to classify input values into two groups.

For applications with a restricted number of stakeholders, SVMs are highly important. The system relies on numerous voice recording sensors to track the voices of patients such as handheld computers, voice recorders, and smartphones. To differentiate between the characteristics of good and unsafe consumers at a higher accuracy rate, SVM is then added to the data (on a cloud-based server). Storage processing method in the machine learning branched into cloud storage and processing and edge storage and processing.