Application of Machine learning and AI for Modern Electrical Engineering

Machine learning (ML) or Artificial Intelligence (AI) is widely applied. The idea of ​​developing algorithms dates back to the early 19th century. However, due to the limitations of computer performance, So it was just a concept of that era. Nowadays, computers are being developed so quickly that computers have sufficient ability to perform calculations. In addition, the development of sensors that can measure various values ​​such as sound, images, temperature and then convert the results into digital values. This data can be stored digitally and can be used to develop ML algorithms.

ML is conceptually different from general algorithms. Normally, an algorithm starts by looking at the relationship between input and output. Then various conditions are created to be consistent with that relationship. Therefore, in collecting data, most experiments are done by adjusting the value of the input of interest and controlling the other inputs to a constant value, and then observing the change in the output. The mathematical relationship between the input and output is evaluated. While the idea of ​​ML is to develop algorithms that can learn and condition themselves from a set of inputs and outputs for example, classifying pictures of dogs. Normally people can easily distinguish pictures of dogs. But when writing a computer program to distinguish images, it is difficult because of complex conditions. In normal development conditions it may be required that a picture of a dog must have 4 legs, 2 eyes, 2 ears and fur.
ML, on the other hand, does not require complex conditions, just a large number of images of dogs to learn. ML is suitable for systems with many inputs and complex relationships with outputs.