Mayank Singh Chauhan's repositories
Steganography
We have implemented a software that can hide messages in images (bmp and png) and audio files.
shake-to-switch
We want people to use gestures to interact with their computers
Starlings-Flock-Simulation
This repository is about modelling of flocking behaviour.
containerapps-albumapi-csharp
Container Apps Quickstart: C# Album API
mayanksingh2298.github.io
My website
extremeText
Library for fast text representation and extreme classification.
i3lock-fancier
Yet another i3lock fork. Now with configuration file!
IITDOmegle
A random chat app based on socket.io
non-convex
Self-contained implementation of non-convex optimization algorithms in python. Steepest Descent, Conjugate Gradient, Newton's Method, Quasi-newton (BFGS), l-BFGS
OpenLinkPrediction
The problem of link prediction has been extensively studied for closed knowledge graphs for quite some time. Researchers have come to with multiple techniques to solve this problem. While closed knowledge graphs are fascinating, much lesser work has been done for open knowledge graphs mostly because of lack of a proper dataset. Recently, [BGWG20] published a dataset - OLPBENCH for standardizing the task of open link prediction. A large part of this thesis is centered around this dataset and trying to improve their baseline algorithms. We try a bunch of adhoc heuristics and find that as simple as they may be - they are good enough to beat the author’s baseline. We also work on a two stage architecture which uses the pre-trained language model [DCLT19] to further improve the link prediction metrics. Not only on open knowledge graphs we find that this architecture also improves the performance of state of the art models on closed knowledge graphs.
pytorch-pretrained-BERT
A PyTorch implementation of Google AI's BERT model provided with Google's pre-trained models, examples and utilities.
shell-windfarm-ai
Optimizing the layout of a wind farm is an interesting and complex optimization problem. The key challenge arises due to the high dimensionality, complex multimodality and discontinuous nature of the search space. Hence, optimizing the layout analytically is impossible. A smarter approach towards solving this problem is through optimization strategies and computer algorithms.