Devendra Patel's repositories
Image-to-Image-Translation-using-CycleGAN
The goal of the image-to-image translation problem is to learn the mapping between an input image and an output image using a training set of aligned image pairs
End-to-end-Masked-Language-Modeling-with-BERT
This example shows you how to start from zero when creating a BERT model, train it on the masked language modeling problem, and then refine it further on a sentiment classification challenge.
Cricket-Data-Analytics
Developed comprehensive player selection criteria, performed data cleansing, transformation, and modeling using Python Pandas and Power Query, and designed an interactive Power BI dashboard to derive strategic insights for crafting the optimal T-20 World Cup 2022 playing 11.
Hindi-English-Translation-
Hindi to English Translation using Seq-2-Seq models
Video-Recommendation-using-Sentiment-Analysis
Short video recommendation from locally stored videos, using facial recognition and rule based mapping for classifying into classes.
Personalized-Cancer-Diagnosis
Using classical machine learning techniques for classifying the data into 9 classes which can be further used for cancer detection.
Recommender-System-using-GNN
Movie Recommender System using GNNs on MovieLens Dataset
Image-Classification-using-GNN
Superpixel Image Classification using GNNs
Self-Supervised-Fake-News-Detection-with-Multimodal-Test-Time-Adapatation-Setting.
Improved the state-of-the-art model accuracy in COSMOS paper by 4%. Implemented a novel architecture that includes global features and local object features of the image and Text features for the self-supervised training using contrastive loss. ”Inspired-Test-Time-Adaptation” to adapt the model to data of different distributions during testing.