Papers I Read Notes and Summaries

CURL - Contrastive Unsupervised Representations for Reinforcement Learning

Introduction

  • The paper proposes a contrastive learning approach, called CURL, for performing...


Competitive Training of Mixtures of Independent Deep Generative Models

Introduction

  • The paper proposes a Competitive training mechanism to train a mixture...


What Does Classifying More Than 10,000 Image Categories Tell Us?

  • The paper is among the first to study image classification at a large...


mixup - Beyond Empirical Risk Minimization

Introduction

  • The paper proposes a simple and dataset-agnostic data augmentation mechanism called...


ELECTRA - Pre-training Text Encoders as Discriminators Rather Than Generators

Introduction

  • Masked Language Modeling (MLM) is a common technique for pre-training language-based...


Gradient based sample selection for online continual learning

Introduction

  • Use of replay buffer (and rehearsal) is a common technique for...


Your Classifier is Secretly an Energy Based Model and You Should Treat it Like One

Introduction

  • The paper proposed a framework for joint modeling of labels and...


Massively Multilingual Neural Machine Translation in the Wild - Findings and Challenges

Introduction

  • The paper proposes to build a universal neural machine translation system...


Observational Overfitting in Reinforcement Learning

Introduction

  • The paper studies observational overfitting: The phenomenon where an agent overfits...


Rapid Learning or Feature Reuse? Towards Understanding the Effectiveness of MAML

Introduction

  • The paper investigated two possible reasons behind the usefulness of MAML...