Tensorflow, Deep Learning and Modern RNN Architectures - Without a PhD

Explore the latest advancements in deep learning and neural network architectures in this comprehensive conference talk. Delve into cutting-edge developments in image recognition, natural language processing, and reinforcement learning. Learn about RNN cells, training replication, multiple layers, and recurrent neural networks. Discover various recurrent cell architectures and their applications in language modeling. Examine the properties of embed-encode techniques and their use in classification tasks. Investigate bidirectional networks and their implementation in toxicity detection. Gain insights into Google Translate's architecture and the role of attention mechanisms. Explore embedding words and the Sequence to Sequence API. Acquire practical tips, engineering best practices, and guidance for applying these advanced techniques in your own projects, all presented in an accessible manner that doesn't require a PhD.

Intro
RNN cell
Training
Replication
Multiple layers
Recurrent neural networks
Recurrent cell architectures
Language model
Shakespeare
Perspective
Embed encode
Embed encode properties
Classification
Bidirectional networks
Building a toxicity detector
Embedding
Encoding
Bidirectional
Toxicity
Google Translate
Adding attention
Embedding words
Demonstration
Sequence to Sequence API