stl10

Official PyTorch implementation of the paper "Self-Supervised Relational Reasoning for Representation Learning", NeurIPS 2020 Spotlight.

[NeurIPS 2021] Why Spectral Normalization Stabilizes GANs: Analysis and Improvements

Steering Self-Supervised Feature Learning Beyond Local Pixel Statistics. In CVPR, 2020.

Implementation of Pre-text invariant representation learning algorithm in pytorch

Pytorch model for classification

A clean, from-scratch PyTorch implementation of I-JEPA trained on STL-10. Built to benchmark representation-space vs. pixel-space prediction against MAE.

SimCLR implementation in PyTorch.

Code for my Master's thesis on Multi‑Task Self‑Supervised Learning for label‑efficient learning. Modular PyTorch framework combining contrastive + pretext tasks with dynamic loss weighting, and centralized/federated training (HAR/STL‑10) to learn compact, robust representations.

Transfer learning from imagenet VGG16 CNN for classifying stl10 database using Keras

GAN implementation in PyTorch

A few machine learning projects that I made using PyTorch

Experiments with Variational Autoencoders in pytorch

CSE-164-Computer-Vision

In this project, we executed an optimized architecture for image recognition on the CIFAR-10 dataset using the Particle Swarm Optimization (PSO) method.

Comparing different learning paradigms on the STL 10 dataset and carrying further analysis in each method

Reconstructing STL-10 images from 9 shuffled, border-eroded patches with an end-to-end network: Sinkhorn permutation layer, differentiable assembly and U-Net inpainting. MAE 0.043 vs 0.182 baseline, 3.26M params, Keras 3.

Reconstructing a 96×96 image from 9 shuffled, border-eroded patches with deep learning (TensorFlow/Keras, STL-10).

Code repository for the paper titled "MIO : Mutual Information Optimization using Self-Supervised Binary Contrastive Learning"

A PyTorch implementation of Masked Autoencoders (MAE) on the STL-10 dataset, demonstrating how self-supervised learning significantly improves classification accuracy and convergence speed on limited labeled data.