Computer Vision
A set of tasks that aim to gain a high level understanding of images or video. Typical tasks include image recognition, object detection, pose estimation and much more.
See more
How the Vision Transformer (ViT) works in 10 minutes: an image is worth 16x16 words
In this article you will learn how the vision transformer works for image classification problems. We distill all the important details you need to grasp along with reasons it can work very well given enough data for pretraining.
Transformers in computer vision: ViT architectures, tips, tricks and improvements
Learn all there is to know about transformer architectures in computer vision, aka ViT.
The theory behind Latent Variable Models: formulating a Variational Autoencoder
Explaining the mathematics behind generative learning and latent variable models and how Variational Autoencoders (VAE) were formulated (code included)
How diffusion models work: the math from scratch
A deep dive into the mathematics and the intuition of diffusion models. Learn how the diffusion process is formulated, how we can guide the diffusion, the main principle behind stable diffusion, and their connections to score-based models.
Generative Adversarial Networks (GANs)
GANs are constructed by two neural networks that compete against each other in a adversarial game, and are proven to be ideal for generating novel data.
See more
GANs in computer vision - Introduction to generative learning
The first article of the GANs in computer vision series - an introduction to generative learning, adversarial learning, gan training algorithm, conditional image generation, mode collapse, mutual information
GANs in computer vision - Improved training with Wasserstein distance, game theory control and progressively growing schemes
The third article-series of GAN in computer vision - we encounter some of the most advanced training concepts such as Wasserstein distance, adopt a game theory aspect in the training of GAN, and study the incremental/progressive generative training to reach a megapixel resolution.
Graph Neural Networks
GNNs are able to extract features from graphs and produce invaluable insights
See more
How Graph Neural Networks (GNN) work: introduction to graph convolutions from scratch
Start with Graph Neural Networks from zero and implement a graph convolutional layer in Pytorch
Best Graph Neural Network architectures: GCN, GAT, MPNN and more
Explore the most popular gnn architectures such as gcn, gat, mpnn, graphsage and temporal graph networks
In-layer normalization techniques for training very deep neural networks
How can we efficiently train very deep neural network architectures? What are the best in-layer normalization options? We gathered all you need about normalization in transformers, recurrent neural nets, convolutional neural networks.
Explainable AI (XAI): A survey of recents methods, applications and frameworks
What is Explainable Artificial Intelligence (XAI), what are the most popular methods, where and how can it be applied
Medical
Deep Learning can also be applied in healthcare and medical applications to solve problems such as diagnosis, prognosis and cure. Understanding medical images is a big part of that endeavour
See more
Understanding coordinate systems and DICOM for deep learning medical image analysis
Multiple introductory concepts regarding deep learning in medical imaging, such as coordinate system and dicom data extraction from the machine learning perspective.
Introduction to 3D medical imaging for machine learning: preprocessing and augmentations
Learn how to apply 3D transformations for medical image preprocessing and augmentation, to setup your awesome deep learning pipeline
Natural Language Processing
An area of Computer Science that focuses on processing and modeling Language. The most popular examples are language translation, voice recognition and text generation.
See more
Why multi-head self attention works: math, intuitions and 10+1 hidden insights
Learn everything there is to know about the attention mechanisms of the infamous transformer, through 10+1 hidden insights and observations
How Positional Embeddings work in Self-Attention (code in Pytorch)
Understand how positional embeddings emerged and how we use the inside self-attention to model highly structured data such as images
Reinforcement Learning
Reinforcement learning is an area of Machine Learning that is about taking suitable action to maximize reward in a particular situation. It has been widely used in solving games but has also numerous applications in real problems.
See more
The idea behind Actor-Critics and how A2C and A3C improve them
Actor critics, A2C, A3C
Unravel Policy Gradients and REINFORCE
Explore Policy-based methods and dive into policy gradients
Unsupervised Learning
Unsupervised Learning is a research field where model are trained without labeled data
See more
Grokking self-supervised (representation) learning: how it works in computer vision and why
A general perspective on understanding self-supervised representation learning methods.
Self-supervised learning tutorial: Implementing SimCLR with pytorch lightning
Learn how to implement the infamous contrastive self-supervised learning method called SimCLR. Step by step implementation in PyTorch and PyTorch-lightning