Advanced Deep Learning with Keras
| Course Details | ||||||||
|---|---|---|---|---|---|---|---|---|
|
||||||||
| Languages | ||||||||
|
||||||||
| Outcomes | ||||||||
|
||||||||
| Price & Terms | ||||||||
|
||||||||
| Technical | ||||||||
|
||||||||
| Specification | ||||||||
|
||||||||
Build Keras functional API networks with several inputs, shared layers and multiple outputs, using basketball data in interactive Python exercises.
Neural networks with complex data flows are the subject here: models that take two, three or more inputs, reuse the same layer for different inputs, or return several predictions at once, all built with the Keras functional API. Working with basketball tournament and regular-season data, you define networks layer by layer, represent teams with categorical embeddings, join inputs with merge layers, inspect a model’s parameters and topology, and finish with one network that tackles a regression and a classification problem together.
What You Will Achieve
- Build, compile and fit a neural network with the Keras functional API, then use it to make predictions
- Turn a high-cardinality column such as team IDs into learned embeddings and reuse one layer across two inputs
- Read a Keras model summary and plot to check a network’s parameters and topology
- Train one network that predicts several targets, or a regression and a classification target at once, and score it with more than one metric
Program Overview
Keras Functional API Basics
Covers the building blocks of a functional network and the full cycle of compiling, fitting, evaluating and predicting.
- Input, dense and output layers
- Building, compiling and visualizing a model
- Fitting on tournament basketball data and evaluating on a test set
Multiple Inputs, Embeddings, and Shared Layers
Moves to networks with two and then three inputs, using embeddings, shared and merge layers, and Keras tools for inspecting a model.
- Category embeddings for high-cardinality team data
- Shared layers and merge layers
- Extending a two-input model to three inputs
- Model summaries, model plots and stacking models
Multiple Outputs for Regression and Classification
Ends with networks that predict more than one target, including a single model that handles regression and classification simultaneously.
- Two-output models for multiple regression targets
- One model for classification and regression
- Evaluating on new data with two metrics
Certificate Awarded (Paid)
Job Outlook
These skills apply to deep learning work on data science teams that build models in Python with Keras, especially models that combine several inputs or predict more than one value.
Good for: Python users who have trained basic Keras models and now want networks with several inputs, shared layers or more than one output.
Keep in mind: The worked examples use basketball tournament and regular-season team data, so practice stays within one sports prediction setting.
Bottom line: An exercise-led run through the Keras functional API that ends with a single network solving a regression and a classification problem together.
Frequently Asked Questions
Which basketball data do the models train on?
The exercises work with basketball data: regular-season games for training and tournament games for testing. Teams are looked up through an embedding, a home-versus-away column becomes a third input, and one model's predictions are later added back into the tournament data.
What does the Keras functional API mean in this course?
It is the way the course defines a model: input layers feed dense layers, which feed output layers, and the model is built from its inputs and outputs. That style is what lets later chapters add second and third inputs, shared layers and extra outputs.
What does stacking models involve here?
Predictions from an earlier model are added as a new column to the tournament data. A second model then takes an input layer with multiple columns, including those predictions, and is fit and evaluated on that data.
User Reviews
Write your review
You must be logged in to post a review.
Reviews can only be posted by registered learners. Log in or create a free account, then you will return here to post.
There are no reviews yet.