This two-day course provides a comprehensive introduction to practical deep learning using MATLAB®. Attendees will learn how to create, train, and evaluate different kinds of deep neural networks. The instructor-led training uses NVIDIA GPUs to accelerate network training.
- Importing image and sequence data
- Using convolutional neural networks for image classification, regression, and other image applications
- Using long short-term memory networks for sequence classification and forecasting
- Modifying common network architectures to solve custom problems
- Improving the performance of a network by modifying training options
Deep Learning with MATLAB is endorsed by NVIDIA's Deep Learning Institute. The Deep Learning Institute offers specialized training also powered by GPUs. Check out their industry-specific content and advanced CUDA programming courses.
Day 1 of 2
Transfer Learning for Image Classification
Objective: Perform image classification using pretrained networks. Use transfer learning to train customized classification networks.
- Pretrained networks
- Image datastores
- Transfer learning
- Network evaluation
Interpreting Network Behavior
Objective: Gain insight into how a network is operating by visualizing image data as it passes through the network. Apply this technique to different kinds of images.
- Feature extraction for machine learning
Objective: Build convolutional networks from scratch. Understand how information is passed between network layers and how different types of layers work.
- Training from scratch
- Neural networks
- Convolution layers and filters
Training a Network
Objective: Understand how training algorithms work. Set training options to monitor and control training.
- Network training
- Training progress plots
Day 2 of 2
Improving Network Performance
Objective: Choose and implement modifications to training algorithm options, network architecture, or training data to improve network performance.
- Training options
- Directed acyclic graphs
- Augmented datastores
Performing Image Regression
Objective: Create convolutional networks that can predict continuous numeric responses.
- Transfer learning for regression
- Evaluation metrics for regression networks
Using Deep Learning for Computer Vision
Objective: Train networks to locate and label specific objects within images.
- Image application workflow
- Object detection
Classifying Sequence Data
Objective: Build and train networks to perform classification on ordered sequences of data, such as time series or sensor data.
- Long short-term memory networks
- Sequence classification
- Sequence preprocessing
- Categorical sequences
Generating Sequences of Output
Objective: Use recurrent networks to create sequences of predictions.
- Sequence to sequence classification
- Sequence forecasting