TAYLLORCOX > Akademie > SAS > Deep Learning Using SAS(R) Software

Deep Learning Using SAS(R) Software

This course introduces the pivotal components of deep learning. You learn how to build deep feedforward, convolutional, recurrent networks, and variants of denoising autoencoders. The neural networks are used to solve problems that include traditional classification, image classification, and sequence-dependent outcomes. The course contains a healthy mix of theory and application. Hands-on demonstration and practice problems are included to reinforce key concepts. Hyperparameter search methods are described and demonstrated to find an optimal set of deep learning models. Transfer learning is covered because the emergence of this field has shown promise in deep learning. Lastly, you learn how to customize a SAS deep learning model to research new areas of deep learning.

Termíny
Target group
Course structure
Prerequisites

Virtual Training nebo e-Learning?

Máme dostatečnou flexibilitu, takže vybírat můžete jak prezenční termíny, tak online kurzy.

Zkuste živý kurz virtuálně

Target group

Machine learners and those interested in deep learning, computer vision, or natural language processing

Course structure

Introduction to Deep Learning

Introduction to neural networks.
Introduction to deep learning.
ADAM optimization.
Dropout.
Batch normalization.
Autoencoders.
Building level-specific autoencoders (self-study).

Convolutional Neural Networks

Applications.
Input layers.
Convolutional layers.
Padding.
Pooling layers.
Traditional layers.
Types of skip-layer connections.
Image pre-processing and data enrichment.
Training convolutional neural networks.

Recurrent Neural Networks

Introduction.
Recurrent neural networks overview.
Sub-types of recurrent neural networks.
Time series analysis using recurrent neural networks.
Sentiment analysis using recurrent neural networks.

Tuning a Neural Network

Selecting hyperparameters.
Hyperband.

Additional Topics

Types of transfer learning.
Transfer learning basics.
Transfer learning strategies.
Transfer learning with unsupervised pretraining.
Customizations with FCMP.

Prerequisites

Before attending this course, you should have at least an introductory-level familiarity with basic neural network modeling ideas. You can gain this neural network modeling knowledge by completing either the Neural Networks: Essentials or Neural Network Modeling course. Previous SAS software experience is helpful but not required.