China
Africa
Explore chapters and articles related to this topic
Time Series Forecasting Using Support Vector Machines
Published in Dinesh C. S. Bisht, Mangey Ram, Recent Advances in Time Series Forecasting, 2021
Generally we face a common problem in machine learning, where we have to select a comprehensive model from the given finite data set. There are subsequent problems of over-fitting and the resulting model ending up with a strongly customized version of a previous model that creates complications. Structural risk minimization (SRM) is basically reducing the complexity of the model by stabilizing the training data. The level of difficulty of the class functions which carried out categorization or regression and the algorithm’s generalizibility are related. There is always a finite set of observations in practical problems. One of the major disadvantages of ERM is that expected risk is not always minimized, which increases the experimental risk over F. Therefore, Vapnik-Chervonenkis (VC) (Vapnik & Chervonenkis, 1971) developed SRM to minimize the expected risk. The technique developed by Vapnik is similar to minimum assumption approaches. The method proposed by Vapnik and Chervonenkis (1971) (VC) gives a universal measure of difficulty and establishes limits on errors as a function of complexity. SRM is the minimization of these limits, which is based on empirical risk and capability of the class of functions.
Application of support vector regression analysis to estimate total organic carbon content of Cambay shale in Cambay basin, India – a case study
Published in Petroleum Science and Technology, 2019
Sanjukta De, Vishal Kumar Vikram, Debashish Sengupta
Support vector machine is a supervised machine learning algorithm used for both classification and regression problems. Support vector machine when used for regression problems, it is known as support vector regression (SVR). It utilizes the structural risk minimization (SRM) principle for classification and regression. The detailed illustration of SVR can be obtained from earlier works (Vapnik 1998; Smola and SchÖlkopf 2004).