Introduction to Deep Learning for Real World Solutions
2-Day Workshop | Next course: coming soon
Price: $800 (Full-time students $500)
Deep learning (DL) is a field of artificial intelligence (AI) in which computer algorithms “learn” to perform advanced pattern recognition on data using neural networks. In today’s world, DL has strong use cases in a wide variety of applications, especially in complex tasks such as classifying images from satellites or predicting medical events in advance.
In this 2-day practical introduction to DL, you will get hands-on experience with the core set of DL algorithms. Through a series of core concept sessions and practical tasks, you will develop an understanding of how DL algorithms function and how to apply them.
The course first provides an introduction to key concepts in AI, including identifying the strengths of DL and where it is best applied to solve a problem. Technical sessions focus on unpacking the different types of DL architectures and the mathematical concepts you need to understand how DL works. We will then work with you to develop practical knowledge of using frameworks and libraries to build DL applications. Focusing on image classification, you will work with data on the cloud to build your DL model, train it, and evaluate its success.
At the end of this course, you will have a solid foundation of core DL concepts on which to continue building your expertise and an intuition for how DL can be applied to solve for real-world problems.
Prerequisites: You must have an understanding of Python to successfully complete this course
Key skills participants will learn from this course:
- Navigating the core Python libraries and DL frameworks
- Understanding the neural network architectures
- Building and training your models on cloud
- Understanding how to fine tune neural networks
- Evaluating the success of your solution
- Applications of artificial intelligence
- What is artificial intelligence, machine learning, and deep learning
- The key skills to effectively use DL on your problem
The three main modes of learning: supervised, unsupervised, and reinforcement learning
Deeper dive into supervised learning
Technical concepts of neural network architectures
- Core Python libraries for DL, including Numpy, Matplotlib, Scikit-Learn and Pandas
- Familiarisation with established DL frameworks, including TensorFlow and PyTorch
- Understanding what your metrics mean and how to finetune
- Consolidation of these concepts with a Jupyter notebook example
- Refresher of matrix operations and logarithms for DL
- High-level walk-through of how functions such as backpropagation, gradient descent, and activation functions make DL possible
- Get set up on Amazon Web Services cloud
- Gain familiarisation with the DL development platform, including best practice for coordinating model build and deployment
- Apply the fast.ai library to successfully build a Convolutional Neural Network (CNN) for image classification. By the end of this final challenge, you will have built a DL solution from start to finish!
Sean did his PhD in robotic vision at the Australian Centre for Robotic Vision, QUT, and holds a Bachelor’s degree in mechatronics. Sean’s PhD focused on training deep convolutional neural networks (CNNs) for autonomous hazard detection on construction sites. Sean has presented his research in both domestic and international conferences and taught numerous workshops on AI. He also acts a community mentor, teaching beginners and advanced deep learning courses.
Seun is a PhD candidate in the Smart Machines Group, UQ, where he focuses on developing artificial intelligence-driven predictive maintenance for complex systems. He is also part of the Asset Management Group at the University of Cambridge. He also holds a Master of Science from Carnegie Mellon University and has over three years’ experience as an academic tutor in engineering and technology courses.