Computer Related
Here is a list of some of my public projects. Feel free to collaborate with me or create your own branches on Github
Machine Learning Projects
Using Convolutional Neural Networks for image classification using the CIFAR10 database
github.com/Pi5hvi/ML_Basic_conv_net_for_cifar10
Using Neural Nets to predict MHC - Peptide binding strength
Every nucleated cell in the human body samples peptides from the pool of proteins being naturally degraded, and presents them on the surface to CD8+ T-cells. The CD8+ T-cells are essentially killer cells, capable of neutralizing various viruses or cancer cells. This presentation process is achieved using the Major Histocompatibility Complex molecules, MHC-I. The peptide-MHC complexes may or may not bind to a T-cell receptor depending on the T-cell’s hyper-variable regions, which allow the different T-cells to recognize different peptides.
This project investigates the probability that any given peptide binds to the MHC to be presented to the T-Cells. We do this by training the neural network to predict factors such as IC-50 and half-life, which are good indicators of the binding strength, given a peptide a MHC as an input.
github.com/Pi5hvi/ML_peptide_MHC_binding
Patrick, a new kind of gradient based optimizer based on ADAM (Work in progress)
This machine learning optimizer, based off adam (for getting bias corrected moment and variance estimates), uses simulated annealing to choose the step size it should take while updating parameters. Initially, it starts with a large step size and over time, the probability that a larger step size is chosen for an update is reduced based on the number of iterations that have occurred.
https://github.com/Pi5hvi/patrick
Machine Learning on Phones(Work in progress)
The idea here is to use Neural Networks to try and predict what upcoming phones may look like. This is a work in progress. I have currently built a python web scraper that scrapes relevant information about every phone available on the website gsmarena.com to build a complete database. The code for the scraper is available on the git project page
https://github.com/Pi5hvi/ML_phones
Optimization Problems
An implementation of the simplex method in MATLAB
github.com/Pi5hvi/MATLAB_Simple_Simplex_example
A basic optimization of the Rosenbrock function using first and second order methods in both MATLAB and python
github.com/Pi5hvi/Rosenbrock_Optimization
MATLAB optimization with a view of iteration history
github.com/Pi5hvi/MATLAB_Optimization_with_iteration_history
MATLAB optimization of a heating element, an example of constrained optimization
https://github.com/Pi5hvi/MATLAB_Optimization_of_heating_element
DFEM in MATLAB
Analysis of a bar element, an example of using MATLAB for Discrete Finite Element Modeling
Various Geometric Computation examples
Implementing basic operations like plotting lines, circles etc. using MATLAB
github.com/Pi5hvi/MATLAB_Basic_Geometric_Operations
Plotting Bezier Curves in MATLAB
github.com/Pi5hvi/MATLAB_Bezier_Curve_Plotting
A python project using "processing" software to demonstrate basic properties of various geometric shapes
github.com/Pi5hvi/Project-Balloon
Plotting a mesh in MATLAB using the marching cubes algorithm on point cloud data
Point cloud data obtained from hardware such as Kinect cameras is basically a mapping of the 3D space in front of the camera in terms of the relative positions of points from which light is reflected toward the camera. We use this data to plot a geometric mesh of the 3D environment in MATLAB using the marching cubes algorithm, implemented from scratch.