SOAI Reading Sessions

Contains notebooks and description files of the article read during our weekly friday sessions.

Want to join us ? Join this discord =)

Reading Session #1: AlexNet

Friday 24/12/2021 and 31/12/2021 were dedicated to read ImageNet Classification with Deep Convolutional Neural Networks article. The files concerning these sessions are in ./AlexNet.

AlexNet is the first large CNN that achieved a state-of-the-art accuracy in 2012 on the large dataset "ImageNet". During this session, we went over:

• How to create a simple MLP and train it on the MNIST dataset
• Create a CNN to avoid overfitting MLPs on large dataset
• Use a pretrained Alexnet on an image
• Summary of the paper

Reading session #2 : Spiking neural networks

we have read Survey of Purpose of Hardware Implementation of Spiking Neural Networks.

The files concerning these sessions are in ./SNN .

• spiking CNN with 98% accuracy on mnist using spikingjelly - Amine Ziad Ounnoughene

Reading session #3 : "TABULAR DATA: DEEP LEARNING IS NOT ALL YOU NEED"

we have downloaded the data from here link
we have read TABULAR DATA: DEEP LEARNING IS NOT ALL YOU NEED.

The files concerning this session are in ./DL IS NOT ALL YOU NEED .

• XGB boost and TabNet trained with the used hyperparameters in the paper on the Covertype dataset
• Challenge : A weighted ensemble average of the XGB boost and the TabNet predictions