Explore

Custom Models, Layers, and Loss Functions with TensorFlow

Approx. 31 hours to complete

Save Course

Go to Course

Course Summary

This course teaches you how to create custom models, layers, and loss functions using TensorFlow, an open-source software library for dataflow and differentiable programming across a range of tasks.

Key Learning Points

Learn how to create advanced custom models with TensorFlow
Discover how to create custom layers and loss functions
Gain hands-on experience with practical coding exercises

Learning Outcomes

Create custom models, layers, and loss functions with TensorFlow
Apply your knowledge to real-world problems
Gain hands-on experience with practical coding exercises

Prerequisites or good to have knowledge before taking this course

Basic knowledge of Python
Experience with machine learning concepts

Course Difficulty Level

Intermediate

Course Format

Online
Self-paced
Practical coding exercises
Video lectures

Similar Courses

Advanced Machine Learning with TensorFlow on Google Cloud Platform
Applied AI with DeepLearning
TensorFlow: Data and Deployment

Related Education Paths

Related Books

Description

In this course, you will:

• Compare Functional and Sequential APIs, discover new models you can build with the Functional API, and build a model that produces multiple outputs including a Siamese network. • Build custom loss functions (including the contrastive loss function used in a Siamese network) in order to measure how well a model is doing and help your neural network learn from training data. • Build off of existing standard layers to create custom layers for your models, customize a network layer with a lambda layer, understand the differences between them, learn what makes up a custom layer, and explore activation functions. • Build off of existing models to add custom functionality, learn how to define your own custom class instead of using the Functional or Sequential APIs, build models that can be inherited from the TensorFlow Model class, and build a residual network (ResNet) through defining a custom model class. The DeepLearning.AI TensorFlow: Advanced Techniques Specialization introduces the features of TensorFlow that provide learners with more control over their model architecture and tools that help them create and train advanced ML models. This Specialization is for early and mid-career software and machine learning engineers with a foundational understanding of TensorFlow who are looking to expand their knowledge and skill set by learning advanced TensorFlow features to build powerful models.

Outline

Functional APIs
A conversation with Andrew Ng: Overview of the specialization
A conversation with Andrew Ng: Overview of course 1
Welcome to the course
Introduction to the Functional APIs
Declaring and stacking layers
Branching models
Creating a Multi-Output model
Multi-Output code walkthrough
Siamese network: a Multiple-Input model
Coding a Multi-Input Siamese network
Siamese network code walkthrough
Connect with your mentors and fellow learners on Slack!
Learn more about the Inception Model Architecture
Energy efficiency dataset
References about the Siamese network
Reference "The distance between two vectors"
Functional API

Custom Loss Functions
Welcome to Week 2
Creating a custom loss function
Coding the Huber Loss function
Huber Loss code walkthrough
Adding hyperparameters to custom loss functions
Turning loss functions into classes
Huber Object Loss code walkthrough
Contrastive Loss
Coding Contrastive Loss
Huber Loss reference
Reference: Dimensionality reduction by Learning an Invariant Mapping
Custom Loss

Custom Layers
Intro custom layers
Introduction to Lambda Layers
Custom Functions from Lambda Layers
Exploring custom Relu with Lambda Layers
Architecture of a Custom Layer
Coding your own custom Dense Layer
Training a neural network with your Custom Layer
Custom Layer code walkthrough
Activating your Custom Layer
Custom Layer with activation code walkthrough
Custom Layers

Custom Models
Intro to custom models
Complex architectures with the Functional API
Coding a Wide and Deep model
Using the Model class to simplify architectures
Understanding Residual networks
Coding a Residual network with the Model class
ResNet code walkthrough
Residual networks lectures (optional)
Custom Models

Bonus Content - Callbacks
Built-in Callbacks
Custom Callbacks
Custom Callbacks code walkthrough
TensorBoard visualization toolkit
References
Acknowledgments

Summary of User Reviews

Learn how to create custom models, layers, and loss functions with TensorFlow. Highly recommended course for anyone interested in improving their understanding of TensorFlow.

Key Aspect Users Liked About This Course

The course covers a lot of ground and provides a comprehensive overview of TensorFlow.

Pros from User Reviews

The course is well-structured and easy to follow.
The instructors are knowledgeable and engaging.
The hands-on exercises are helpful for reinforcing concepts.
The course provides a good balance of theory and practical application.
The course is suitable for both beginners and advanced learners.

Cons from User Reviews

The course can be challenging at times, especially for those new to TensorFlow.
The pace of the course may be too slow or too fast for some learners.
Some of the coding exercises can be time-consuming and may require a lot of debugging.
The course does not cover some advanced topics in depth.
The course may not be suitable for those looking for a quick introduction to TensorFlow.

Recommended for you