Abstract: Come and learn how the TensorFlow APIs work for a variety of scenarios from Computer Vision to Sequence Modelling and Natural Language Processing. Ill start with a high level overview of what coding the simplest possible neural network looks like a hello, world of AI if you will.
From there, I will look at extending this for Deep Neural Networks where multiple neurons operate together in layers to learn concepts like the contents of simple images or interpreting patterns to make predictions.
After that, well see how to extend the theory for Computer Vision, with the idea of convolutions learnable filters that can be used to break down images so that computers recognize them.
Well spend most of our time with Natural Language Processing and how to train models with NLP techniques. Well see how computers break down words into embeddings or higher dimensional vectors, whose direction can be used to establish sentiment. With this, youll then see how a computer can begin to understand the meaning of text and well see how to train AI models to detect things like whether a movie review was positive or not. The techniques can then be extended to text prediction which leads to text generation and youll learn how to create a simple AI model that creates its own text. Well use a simple example of one that uses traditional Irish music, and which generates lyrics for songs based on your desired input. The same work was used by Google and the folks from the Stargate TV show to create the famous 'Stargate AI' experiment [https://www.thecompanion.app/ai-project/], where the cast and crew of the show reunited for a table read of a script entirely generated by an AI!
Bio: Laurence Moroney leads AI Advocacy at Google, working with the Google AI Research and product development teams. He's the best-selling author of 'AI and Machine Learning for Coders,' as well as the instructor on the Fundamentals of TinyML course at HarvardX, and the popular TensorFlow specializations with deeplearning.ai and Coursera. He's passionate about empowering software developers to succeed in Machine Learning, democratizing AI as a result. Laurence is based on Washington State in the USA.