Humza Iqbal

I am a student at UC Berkeley studying Computer Science with particular interest in Machine Learning, Data Science, and Natural Language Processing.

Experience

A collection of my research and work experiences.

Data Science Intern at HealthPals

May 2016 - Aug 2016
Using sensitivity analysis to understand how different risk factors interact with each other to contribute to heart disease and implemented analytics tool for patients to view 10 year risk of heart attack

Undergraduate Research at Machine Learning at Berkeley(MLaB)

May 2016 - Present
Working on Machine Learning project with Grand Rounds in order to find anomalies in medicare data as well as general analysis

Machine Learning Intern at Elastica

May 2015 - Aug 2015
Implemented machine learning classifiers for NLP problems with 90% test accuracy along with a data pipeline

Strategy Lead and Electrical Member for UC Berkeley Solar Car Team (CalSol)

Feb 2015 - Aug 2015
Used Machine learning in conjunction with race data to figure out optimal route to race car on a closed circuit track and wrote microcontroller code to ensure that the car runs safely and efficiently

Projects

Artificial Neural Network

Learning Machine capable of both classification and regression

Gitlet

Miniature version of Git version control.

Wordnet

Program to track history of words over time

Maximum Acyclic Subgraph approximator

Approximation algorithms for an NP hard optimization problem

Random Forest

Random Forest learning machine for classification

Artifical Neural Network

Learning machine for classification and regression.

Implemented a neural network in python which can take on an arbitrary number of hidden layers and be used for both regression and classification, along with unsupervised learning."

Neural Network Code

Gitlet

Miniature version Git

Implemented my own version of Git where in the user can perform most of the basic functionality of Git incluing commit, merge, checkout, and rebase."

Gitlet Code

Wordnet

Historian of the written word

"I use the Google NGrams corpus along with custom data structure to track the usage of various words across known literature dating back to the 1400s. In addition, I also provide an empirical representation of Zipf's law"

Wordnet Code

Acyclic Subgraph approximator

Approximation algorithms for an NP hard problem

"I implement an approximator for an NP hard optimization problem which uses multiple different algorithms per instance of the problem and takes which ever output is the highest performing. The algorithms include a genetic algorithm, a reduction to Integer Linear Programming, and randomized algorithms"

Acyclic Subgraph code

Random Forest

Ensemble learning machine

"I implement a Random Forrest using a combination of several decision trees, which are trained to make splits which minimize either the cross or gini entropy"

Random Forest Code