4.7.1: Beyond perceptrons: Convolutional Neural Network (CNNs) in the web browser

Опубликовано 6 февраля 2023, 16:16

In this video learn about a new model architecture known as a Convolutional Neural Network (CNN) that overcomes some of the limitations that a simple Multi Layer Perceptron (MLP) has such as being fooled by position of the object in the image or its size and rotation. CNNs are widely used today across many popular architectures so this is a key evolution of your knowledge when working with image like data in the browser. You will learn about new concepts and terminology such as convolutions, filters (kernels), padding, stride, and max pooling to build a new solution to your MNIST classifier that is more robust than before.

"Catch more episodes from Machine Learning for Web Developers (Web ML) → goo.gle/learn-WebML
Check out TensorFlow on YouTube → goo.gle/TensorFlow-YouTube
Subscribe to Google Developers → goo.gle/developers

Connect with Jason Mayes to ask questions:
LinkedIn → goo.gle/3GwgeLw
Twitter →goo.gle/3Xh6MT7
Discord →goo.gle/3WWVO5t

Use #WebML to share your learnings and creations from this course to meet your peers on social media!

See what others have already made with Web ML → goo.gle/made-with-tfjs"