Neural Networks from 0.0 to Dancer
300 May 20, 2022, 1:10 PM - 2:00 PM
Azure Developer, Active Solution
For many developers, neural networks are considered an area of mystery and magic only fully understood by advanced data scientists and evil government organizations. However, the fundamental concepts of concepts of neural networks are very basic, and these form the building blocks for designing and training more sophisticated networks that can be used in real-world applications.
In this demo-intensive session Alan will begin with showing how a single neuron can be used to predict apartment prices and then extend this basic the model to a full neural network. Image classification will then be used to demonstrate the limitations of linear neural networks, and the concepts and advantages of convolutional neural networks will be explained and demonstrated. Generative adversarial networks (GANs), that can enhance images, perform style transfer and create beautiful art will be covered, with Alan demonstrating how a trained GAN can improve your dancing style from Andy McCluskey to Bruno Mars.
By the end of this session, you will have a solid understanding of neural networks, and understand the difference between a sigmoid and a leaky relu.