دانلود کتاب پردازش سیگنال و یادگیری ماشین با برنامه های کاربردی

Learning has always a goal. For this reason, for a given process, the question is what can we learn about it? Despite the diversities in processes, one can find out the common principles underlying them. Therefore, we discuss stochastic processes from a general point of view, first. As for practical applications, we will consider them through the lens of historical developments by considering the following aspects: 1. The invention of numbers and number systems is our starting point. From numbers and their representations, humans invented the theory of signals and systems. Signals do not occur isolated but within a process. 2. The naturally occurring processes are not deterministic but rather stochastic. Hence, the underlying states are often unknown. That means those processes are mostly described with hidden stochastic models (HSM). 3. Signal processes contain information. However, this information is not directly visible or understandable to humans. The information is hidden in properties of the processes. Properties of signal processes are mostly not directly measurable, and we have to find them out, at least approximately. 4. Information about the unknown elements is necessary. One needs methods to capture them, and here lie the major challenges. To address these challenges, we concentrate on machine learning techniques. Real-world problems are conceptualized, explained in applied domains. The book is intended for persons interested in the techniques and foundations related to signal processing. As mentioned previously, signals occur within processes and are considered not to be fully observable and predictable; hence, they are ruled by probabilities. Therefore, they are termed stochastic processes. We concentrate on properties of signals that are useful for practical applications. One has rather to manipulate the information available about processes using computations. Doing this properly is a major challenge. The computations needed depend very much on the type of signals and the application domain.