Entropy Randomization in Machine Learning

This book offers a fresh perspective on machine learning through entropy randomization, addressing uncertainty in data and models effectively.

The book Entropy Randomization in Machine Learning introduces a novel methodology in the field of machine learning known as entropy randomization. This approach aims to derive optimal solutions amidst uncertainty, specifically when dealing with uncertain data and models related to the subjects of study. By employing randomized machine-learning techniques, the book illustrates how models can incorporate random parameters and utilize maximum entropy estimates of probability density functions under balanced conditions with the available data.

In Entropy Randomization in Machine Learning, the author meticulously derives optimality conditions expressed through nonlinear equations featuring integral components. To tackle these complex equations in a probabilistic context, a new numerical random search method is presented. This method not only enhances the theoretical understanding of randomized machine learning but also lays the groundwork for practical applications across various domains. The book explores several case studies, including binary classification, modeling the dynamics of Earth's population, predicting seasonal electric load fluctuations, and forecasting changes in the thermokarst lakes area in Western Siberia.

Key features of the book include a comprehensive overview of the randomized machine-learning problem, covering everything from data processing to the development of algorithmic procedures. Additionally, it presents innovative numerical methods for random global optimization and the computation of multidimensional integrals, along with a universal algorithm tailored for randomized machine learning. This resource is particularly beneficial for students, researchers, and professionals engaged in artificial intelligence and machine learning, as well as those involved in various forecasting challenges.