Explore deep learning concepts and their implementation in Scala in this conference talk from Scala Days Berlin 2018. Learn about artificial neural networks, computer vision, speech recognition, and machine translation. Understand the fundamentals of deep learning, including tensors, models, classifiers, and the training process involving forward pass, backpropagation, and gradient descent. Discover why deep learning works through composition and narrow intelligence. Examine the benefits of using Scala for deep learning, such as type safety, and compare three deep learning libraries. Follow along with example code to solve a business problem using supervised learning, convolutional neural networks, and model deployment. Gain practical insights into building, training, and deploying deep learning models with Scala.
Overview
Syllabus
Intro
Electricity
Deep Learning
Getting Started
Theres no free lunch
Deep Learning in 3 Steps
What is Deep Learning
Artificial Neural Networks
Computer Vision
Speech Recognition
Machine Translation
What do these examples have in common
Overlaying the examples
Generic functions
Models
Classifiers
Tensors
Tensor Models
Artificial Neuron
Architecture
Machine Learning
Training Data Examples
How Does Training Work
First Step
Forward Pass
BackPropagation
Gradient Descent
Why Deep Learning Works
Composition
Narrow Intelligence
Conclusion
Why Scala
Type Safety
What features should a deep learning library provide
Three deep learning libraries
Example code
Business problem
Look for Existing Solutions
Supervised Learning
Data Source
Architectures
Convolutional
Neural Network
Model Building
Training
Training with more neurons
convolutional layer
network classifier
run model
deployment
main takeaways
Taught by
Scala Days Conferences
