This project has a comprehensive exploration of two key topics: Softmax Regression and Contrastive Representation Learning.
Softmax Regression, also known as multi-class logistic regression, forms the base of many classification tasks in machine learning. It involves extending binary logistic regression to handle multiple classes, making it an one of the best tools for multi-class classification problems. In this assignment, we are tasked with implementing Softmax Regression using PyTorch, a popular deep learning framework, to classify images from the CIFAR-10 dataset.
Contrastive Representation Learning, on the other hand, is part of unsupervised learn- ing, specifically focusing on learning representations of data in a way that enhances their discriminatory power. By training a neural network to maximize the similarity between similar instances while minimizing the similarity between dissimilar instances, Contrastive Representation Learning aims to create embeddings that capture meaningful relationships within the data. In this project I have built a neural network architecture and train it using a contrastive loss function.
Folder Structure
python run . py \ −−mode name of mode
<other hyper−params here>
