TensorFlow Training on the JVM

Explore TensorFlow training on the Java Virtual Machine in this 34-minute conference talk by Christoph Henkelmann from DIVISIO at MLCon. Learn why and when combining TensorFlow models with Java can be beneficial for commercial projects running on the JVM. Discover TensorFlow basics, including its architecture, operations, and sessions. Understand how TensorFlow training works and follow the step-by-step process of defining an initial graph, running sessions, and debugging. Gain insights into the advantages and challenges of using Java for machine learning tasks, and see practical examples of implementing TensorFlow in a JVM environment.

Introduction
What we want to achieve
Why Java
Java can be nice
Java is not cool
When is this a bad idea
When can it make sense
Why
TensorFlow Basics
TensorFlow Architecture
What is TensorFlow
Operations
Sessions
How TensorFlow training works
How does learning work
Plan of attack
Defining the initial graph
Running the session
amnesty
mirror
initial code
debugging
running the graph