From the course: Computer Vision on the Raspberry Pi 4
Join today to access over 24,800 courses taught by industry experts.
Training neural networks
From the course: Computer Vision on the Raspberry Pi 4
Training neural networks
- [Instructor] To understand neural networks, it helps to view them mathematically. Every neural network embodies an equation that relates outputs to inputs. For example, the network in this slide accepts four inputs, feeds them into three nodes and produces one output. The output can be expressed as a function, determined by the network's weights, inputs, and activation functions. This serves as our model of how the output is determined by the input. Many aspects of the model are selected in advance, such as the network structure and its activation functions. The inputs are determined by the system, so the only way to improve the model is to change the weights. The process of finding the best weights is called training. The most popular method of training a neural network is backpropagation. This requires five steps that are repeated several times. The first step computes the network's output for a set of inputs and…
Download courses and learn on the go
Watch courses on your mobile device without an internet connection. Download courses using your iOS or Android LinkedIn Learning app.