Explore the fundamentals of deep learning using NVIDIA's DIGITS (Deep Learning GPU Training System) in this introductory lesson. Learn to create, test, and optimize neural networks for image classification tasks. Discover how to configure networks, prevent overfitting and underfitting, perform single image classification, and improve performance through data augmentation. Gain insights into network deployment, including rapid classification with Tegra. Access hands-on lab exercises and additional resources to reinforce your understanding of deep learning concepts and practical applications.
NVIDIA Deep Learning: Getting Started with DIGITS - Class 2
Overview
Syllabus
Intro
WHAT IS DEEP LEARNING? A Classification Example
IMAGE CLASSIFICATION WITH DNNS
HOW DO I GET DIGITS? Two ways to install it
TRAINING AND VALIDATION DATA Why do we need it?
NVIDIA DIGITS Network Configuration - Start with a Standard Network
LENET Network Configuration
OVERFITTING AND UNDERFITTING How can I use DIGITS to tell me this is happening?
NVIDIA DIGITS Single Image Classification
SINGLE IMAGE CLASSIFICATION RESULTS
NETWORK PARAMETERS
NVIDIA DIGITS Modifying your Network
USING AN AUGMENTED DATA SET
ANOTHER WAY TO IMPROVE PERFORMANCE Data Augmentation Example augmentation - Inverted copies of the input data
DEPLOYING YOUR NETWORK
DEPLOYMENT WITH TEGRA Rapid Classification Anywhere
NVIDIA DIGITS Resources
HANDS-ON LAB
DEEP LEARNING LAB SERIES SCHEDULE
Taught by
NVIDIA Developer
