rupak-118 / ai-papers

ML/DL papers which I found useful. Lots of them.

AI-papers

This repository contains a list of papers as discussed on my blog series (23 Deep Learning Papers To Get You Started) on Medium.

• Part 1 - https://medium.com/@rupak.thakur/23-deep-learning-papers-to-get-you-started-part-1-308f80d7bba2
• Part 2

It also contains additional research papers which I keep finding useful along my DL journey.

Part 1

1. A Few Useful Things to know about Machine Learning - Pedro Domingos : A very useful paper summarizing all the key learnings from an ML standpoint

2. Introduction to CNNs : An article, pretty different from a conventional paper, but provides a comprehensive explanation through mathematical concepts and derivations of basic CNN elements

3. Visualizing and Understanding CNNs - Zeiler and Fergus : Landmark paper in the history of deep learning. This opened up the doors to building more interpretable and easily explained deep learning models

Part 2

4. Xavier initialization - Glorot, Bengio : Talks about the importance of appropriate weight initialization in neural networks

5. Delving Deep into Rectifiers - Surpassing Human Level Performance on ImageNet Classification : Introduces PReLU activation function and He initialization (another weight initialization scheme)

6. BatchNormalization : Talks about internal covariate shift and accelerating the training process of a deep neural network by reducing it

7. Overview of Gradient Descent Optimization Algorithms : Explains the math and intuition behind different optimization algorithms coming under the Gradient Descent umbrella

8. Dropout - Hinton et al. : Introduces the hugely popular regularization technique, Dropout. Dropout has been a key component in most neural network architectures since, and has also been highly successful in competitions.

ai-papers's Issues

Are you open to contributions?

Hi Rupak
Really like this small selection of fundamental papers on the deep learning framework.
I see that last activity on this repo is >3y ago. Did you manage to work on the second part at some point? I will probably go through all listed articles, and I was wondering how I could contribute to the serie if I ever makes such summaries on my side. Co-authoring a second medium article is an idea I was considering. What do you think?