Neural Structured Learning in TensorFlow

Explore Neural Structured Learning in TensorFlow through this 43-minute conference talk from TF World '19. Discover how this open-source framework enables both novice and advanced developers to train neural networks with structured signals. Learn about the advantages of learning with structure, model robustness, and practical applications in image classification. Gain insights into neural graph learning, adversarial learning, and the use of libraries, tools, and trainers. Delve into hands-on tutorials and explore concepts like image semantic embedding and neural architecture. Presented by Da-Cheng Juan and Sujith Ravi, this talk provides a comprehensive overview of Neural Structured Learning's potential in vision, language understanding, and general prediction tasks.

Intro
How a Typical Neural Net Works
Neural Structured Learning (NSL)
Structure Among Samples
NSL: Advantages of Learning with Structure
Scenario Il: Model Robustness Required Example task: Image Classification
NSL: Neural Graph Learning Joint optimization with label and structured signals
NSL: Neural Graph Learning in Practice
NSL: Adversarial Learning
Libraries, Tools and Trainers
NSL Resource: Hands-on Tutorials
Learning Image Semantic Embedding
Neural Architecture