- Hands-On Neural Networks
- Leonardo De Marchi Laura Mitchell
- 182字
- 2025-04-04 14:15:16
Neural Network Fundamentals
Artificial neural networks (ANNs) are a set of bio-inspired algorithms. In particular, they are loosely inspired by biological brains; exactly like animal brains, ANNs consist of simple units (neurons) connected to each other. In biology, these units are called neurons. They receive, process, and transmit a signal to other neurons, acting like a switch.
The elements of a neural network are quite simple on their own; the complexity and the power of these systems come from the interaction between the elements. A human brain has more than 100 billion neurons and 100 trillion connections.
In the previous chapter, we introduced a supervised learning problem. In this chapter, we will cover the main building blocks to create Neural Networks (NNs) to solve such a problem. We will cover all of the elements to create a feedforward neural network, and we'll explain how to train it, implementing it from scratch and using Keras.
Following are some important topics that we will be covering in this chapter:
- The Perceptron
- A simple Feed forward Neural Network
- FFNN in Python from scratch