[UDM] Deep Learning From Scratch Course (Oct 2024) + Full Videos & Resources

Get into Deep Learning at all Levels
Deep Learning From Scratch Course is a course created to satisfy the needs of all people who desire to learn how deep learning actually works, not through existing libraries, but by building neural networks with absolutely no libraries. You will start at the starting block of what neural networks are based on, which includes forward propagation, backpropagation, and gradient descent, and translate them in a step-by-step manner in Python buffers. It is not only a practical exercise, but this strengthens your understanding of math and programming, so advanced topics can be learned.

An Algorithmic Adventure of the Real Coding
As you progress through the Deep Learning From Scratch Course, you will step by step transition to more complex neural networks such as convolutional neural networks (CNNs). Every step will be filled with coding activities and applied training to strengthen what you have learned. You will write code that implements every network layer all by yourself, none of this shortcut stuff that just lets you get on to the point that machine learning works, so that when your model predicts and trains, you know what is happening behind the scenes.

Learn to master CNNs and Real Image Datasets
Once you have learned the fundamentals, you will enter the world of CNNs (they are very strong models to classify and recognize images). You will discover the basics of deep learning, including dropout, batch normalization, learning rate optimization, and how to process and classify image databases, such as MNIST and CIFAR-10. All the information is presented in simple terms, together with code, hence it was easy to relate theory and practice.

Construct Your Object Detection System
The Deep Learning From Scratch Course ends with an effective practical exercise: training an end-to-end object detection system with your own CNN trained. You will complete the whole procedure of preparing image data and defining a model architecture for training and testing of your model. In the course of the project, you will code a working application and learn a lot about the construction of the modern AI system, without using external deep learning libraries.