Intro to Neural Network
TLDR: This page provided an introduction to neural networks, specifically the perceptron. Traditional programming, where rules are explicitly defined, is contrasted with perceptron programming, where rules are derived from adjusting weights and thresholds by a learning algorithm based on data. A basic case study of a non-negative classifier is used to illustrate this approach, highlighting how predefined mathematical characteristics can be replaced by data-driven inference.
Conclusion
In this article, using basic case study of non-negative classifier, we have seen a different approach of problem solving. Instead of defining rules as in ordinary programming, we use data to inference the rules. Instead of defining mathematical characteristics of the problem, we use learning algorithm to adjust the weights and threshold of the perceptron.
We could generalize the difference between ordinary programming and perceptron programming as follows:
So instead of defining the rules explicitly, we let the computer learn the rules from the data 😎.
This extracted rules then used to solve the problem. Here is the complete diagram of the steps:
References
Hello, Perceptron: An introduction to artificial neural networks
Masterclass-Machine Learning untuk Rekayasa Teknik