In this liveProject, you’ll step into the role of a developer working for car rental company CarsUnlimited. You have a bold new idea for an app that can revolutionize your business. You envisage a service where a customer can snap a picture of any car on the street, and instantly find out if CarsUnlimited has that make and model available. Your colleagues have the app UI covered; your challenge is to build the algorithm to classify the cars that were snapped by the user. You decide to develop a machine learning algorithm for solving this problem.
You’ll start with cleaning your data, before experimenting with classical machine learning algorithms such as SVMs, Random Forests, Decision Trees, and assess their performance metrics. With an entire range of parameters to tune and improve, you will learn about how to automate this search. From there, you’ll dive into building a deep neural network and explore different search strategies to finalize your powerful new neural network architecture.
This project is designed for learning purposes and is not a complete, production-ready application or solution.
This liveProject is for beginners to machine learning who have some existing knowledge of Python and linear algebra. To begin this liveProject, you will need to be familiar with:
- Intermediate Python
- Basics of Jupyter Notebook
- Basics of scikit-learn and Autosklearn
- Basics of TensorFlow, Tensorboard and Keras Tuner
you will learn
In this liveProject, you’ll learn to use machine learning and deep learning algorithms for image classification. You’ll learn to identify what type of machine learning algorithms would prove useful in a given setting and to highlight the importance of experimentation. The skills that you build in developing, debugging, and understanding machine learning decisions are easily transferable to any machine learning project.
- Data visualization to understand the available data
- Importance of data pipelines when training data intensive machine learning algorithms
- Advantages of caching and prefetching to make training pipelines more efficient
- Exploring classical machine learning algorithms as baselines to solve machine learning problems
- Automating exploration of classical machine learning algorithms
- Learning to build neural networks and the need for deeper neural networks
- Comparing machine learning experiments using Tensorboard
- Automating exploration of neural network architectures to search for the best possible architecture for a given dataset