Noise-Contrastive Estimation - CLEARLY EXPLAINED!
Noise-Contrastive Estimation is a loss function that enables learning representations by comparing positive and negative sample pairs. It came into the limelight as a workaround to approximate softmax in NLP but is now being used in a lot of experiments in Self-supervised representation learning. #selfsupervised #representationlearning #noisecontrastiveestimation

▶︎
Probabilistic Programming - FOUNDATIONS & COMPREHENSIVE REVIEW!

▶︎
Rejection Sampling - VISUALLY EXPLAINED with EXAMPLES!

▶︎
MoCo (+ v2): Unsupervised learning in computer vision

▶︎
BYOL: Bootstrap Your Own Latent: A New Approach to Self-Supervised Learning (Paper Explained)

▶︎
Contrastive Loss : Data Science Basics

▶︎
Reparameterization Trick - WHY & BUILDING BLOCKS EXPLAINED!

▶︎
Contrastive Learning with SimCLR | Deep Learning Animated

▶︎
Evidence Lower Bound (ELBO) - CLEARLY EXPLAINED!

▶︎
Contrastive Learning - 5 Minutes with Cyrill

▶︎
SimCLR: A Simple Framework for Contrastive Learning of Visual Representations

▶︎
Transformer Neural Networks Derived from Scratch

▶︎
Kalman Filter - VISUALLY EXPLAINED!

▶︎
The Reparameterization Trick

▶︎
Contrastive Learning in PyTorch - Part 1: Introduction

▶︎
Supervised Contrastive Learning

▶︎
Boltzmann Machine - Explained!

▶︎
SimCLR Explained!

▶︎
Importance Sampling - VISUALLY EXPLAINED with EXAMPLES!

▶︎
C4W4L04 Triplet loss

▶︎
