Advanced CNN
This course build on the subject of Convolutional Neural Networks and dives into the complex pre-trained state- of-the-art CNN architectures. It also gives students a preview of what transfer learning is and why it is such a powerful concept in Deep Learning. By the end of this course students will be able to have implemented and explored pre-trained models such as ResNet, VGG16, and Inception3.
4 hours of instruction
This course build on the subject of Convolutional Neural Networks and dives into the complex pre-trained state- of-the-art CNN architectures. It also gives students a preview of what transfer learning is and why it is such a powerful concept in Deep Learning. By the end of this course students will be able to have implemented and explored pre-trained models such as ResNet, VGG16, and Inception3.
OBJECTIVES
- Define the need for advanced CNNs
- Describe optimization using regularization and apply to CNN
- Implement a VGG16 on the same dataset to compare performance and explore the concept of transfer learning
PREREQUISITES
Convolutional Neural Networks (CNN) for Image Recognition
SYLLABUS & TOPICS COVERED
- Advanced CNN Architectures
- Use cases for advanced CNN architectures
- Difference between CNNs and advanced CNNs
- Baseline CNN
- Implementation of a baseline CNN
- Measuring performance of a baseline CNN
- Advanced CNN Models
- Introduction to popular pre-trained advanced CNN models
- VGG16, Inception, and ResNet architecture and implementation details
- The concept of transfer learning and it use cases
SOFTWARE REQUIREMENTS
You will have access to a Python-based JupyterHub environment for this course. No additional download or installation is required.
About Instructor
