Autoplay
Autocomplete
Previous Lesson
Complete and Continue
Deep Learning 101 with Python and Keras
Welcome to Deep Learning 101!
How does the course work? (4:00)
All resources
Module 1: Introduction to Deep Learning
Quick overview of Deep Learning (18:30)
Questions
Answers to questions
Module 2: Building blocks of Deep Learning
Forward pass in Neural Networks (18:32)
Backward pass in Neural Networks (15:59)
Questions
Answers to questions
Module 3: Exercise - Let's build a simple Neural Network
Setting up a virtual environment (8:19)
An alternative development environment
Importing and preparing the data (10:34)
Building the model (17:48)
Model evaluation (6:50)
Assignment
Module 4: Hyperparameters of Neural Networks
Download your hyperparameters cheat sheet
All hyperparameters overview (7:15)
Determining number of neurons and hidden layers (4:23)
Loss function, optimizers and learning rate (4:47)
Activation function, batch size and # of epochs (4:46)
Weight initialization and regularization (2:20)
How to set up hyperparameters in Keras (12:20)
Questions
Answers to questions
Module 5: Over/underfitting problem and solutions
What is over/underfitting and how do we deal with it? (11:32)
L1/L2 Regularization (8:07)
Dropout Regularization (5:23)
Other techniques: data augmentation, early stopping (5:29)
How to set up each solution in Keras (8:43)
Questions
Answers to questions
Module 6: Unstable Gradients
What is vanishing/exploding gradients problem? (5:54)
Choosing the right initializer (3:40)
Non-saturating activation functions (5:24)
Batch normalization (13:10)
Gradient clipping (6:31)
How to set up each solution in Keras (9:54)
Questions
Answers to questions
Module 7: Computational time
Overview of solutions to long training times of Neural Networks (1:52)
Normalizing your data (3:17)
Using mini-batches (13:56)
Selecting the right optimization algorithm (13:36)
Pruning your network (1:49)
Learning rate scheduling (7:57)
How to set up each solution in Keras (10:52)
Questions
Answers to questions
Module 8: Exercise - Setting up neural networks
Introduction to the data and the notebook (3:44)
Initializing the network (7:14)
Dealing with overfitting (7:12)
Learning rate scheduling (6:16)
Options for initializing the network (5:31)
Assignment
Module 9: Model diagnosis and making improvements
Model evaluation metrics (7:01)
Levels of performance (4:11)
Determining optimal performance level for your model (9:49)
Deciding on improvements to make (3:22)
Questions
Answers to questions
Module 10: Hyperparameter tuning
Hyperparameter tuning problem and common solutions (6:33)
Advanced hyperparameter tuning algorithms (7:19)
Questions
Answers to questions
Module 11: Exercise - Improving the model
Diagnosing the model (4:01)
Hyperparameter tuning (11:57)
Training the model further (2:44)
Assignment
BONUS: RNNs and CNNs
RNNs: Basics (15:45)
RNNs: LSTMS and GRU cells (10:06)
RNNs: Implementation using Keras (6:49)
CNNs: Basics (12:06)
CNNs: Calculating results in practice (14:19)
CNNs: Implementation using Keras (2:42)
Congratulations on finishing the course!
A word from your instructor (0:59)
Let me know what you thought about the course
Continue learning...
Hands-on Data Science: Complete your first portfolio project
What is vanishing/exploding gradients problem?
Lesson content locked
If you're already enrolled,
you'll need to login
.
Enroll in Course to Unlock