Argha Sarkar's repositories
google-sheets-js
An easy to use synchronous google sheets reader without the use of callbacks
exam-papers
Exam papers and exercise sheets
neural-network-in-python
Predict the number of bikes being used given several metrics about any given day. This project implements a neural network and all the functionality required (eg: forward pass, back propagation and optimisation using gradient descent) using just python. No machine learning libraries like tensorflow are used in this project.
a32nx
The A32NX Project is a community driven open source project to create a free Airbus A320neo in Microsoft Flight Simulator that is as close to reality as possible. It aims to enhance the default A320neo by improving the systems depth and functionality to bring it up to payware-level, all for free.
easy-trie
A lightweight & easy-to-use implementation of the Trie data structure written in TypeScript. This can be used for searching for words to autocomplete and also for spell-checking.
AlexaSkillNextTube
An Alexa skill to tell me when the next tube is from my station
AlexaWildyPenguinChecker
Alexa skill for checking if there are any wilderness penguins.
arghasarkar.github.io
Personal web page
autocomplete
IDE-style autocomplete for your existing terminal & shell
aws-mobile-appsync-events-starter-react
GraphQL starter application with Realtime and Offline functionality using AWS AppSync
cs907
Dissertation: An adversarial approach to identity preserving gender obfuscation in facial images
dog-breed-classifier-using-neural-network-with-transfer-learning-on-imagenet
A neural network using Keras to classify dog breeds from images. This is a hard task for humans and even harder for computers. However, using transfer learning by pre-training some convolutional layers on imagenet and then retraining the fully connected layers on the dog breed training images, an exceptionally high level of accuracy has been achieved.
embedded-mysql-plugin
gradle plugin for embedded mysql
face-generation-using-gans
Use GANs with normalization techniques like dropouts, batch normalization along with having a low variance in kernel weight initialization, achieve realistic images of faces trained on the CelebA dataset. Images also have been generated of hand written digits after being trained on the MNIST dataset. This would be useful for generating training images where enough labelled data is unavailable.
first-contributions
🚀✨ Help beginners to contribute to open source projects
Hacktoberfest
Opportunity to start open source and PR experience
high-performance-computing
Optimize thermodynamics simulation using OpenMP
MechDawn
Issue tracking for MechDawn
multi-agent-systems
Intelligent agents in a multi-agent system taking part in different forms of virtual auctions with differing goals and payoffs.
netflixparty-chrome
A tool for synchronized Netflix consumption.
oas-raml-converter
Converts between OAS and RAML API specifications:
tv-script-generation-with-deep-learning
Generate scripts for the Simpsons TV show using deep learning. Using a recurrent neural network with LSTM (Long Short Term Memory) cells along with word embeddings, it is possible to find patterns within the data. Adding a bit of randomness to the probability of which word should be next in the pattern, it is possible to generate TV scripts. Parts of the scripts will not make grammatical sense. It's still a cool project nonetheless.