Stephen C Ashmore

I solve difficult problems, I'm a Software Engineer & Machine Learning expert.
LinkedIn StephenCAshmore

Stephen Ashmore has earned a Master's Degree from the University of Arkansas, and has been a PhD Candidate there working with machine learning and robotics. While his work has focused on machine learning, big data, robotics, and neural networks, he is very competent with full-stack development and back end work. He has been working in Academia for 5 years, with experience on many different back end and full-stack projects.

Skills

Programming [OOP]
C++, Rust, Java, Lua, Python
Deep Learning
Neural Networks, Machine Learning, Data Mining, Big Data, Time Series, Decision Trees, Robotics, Data Science
Web Development
HTML, CSS, Javascript, Ruby, Meteor
Backend
SQL, Django, MongoDB, PHP, NoSQL Databases, Redis
Systems
Linux, Mac OS X, Windows

Experience

University of Arkansas

Smart Robots and Machine Learning Researcher Jul 2015 - May 2017

Researching machine learning and robotics, this position in Dr. Michael Gashler's Predictive Machine Learning and Modeling Laboratory involved creating novel new architectures and algorithms for deep learning and robotics. Resulted in several publications relating to robotics, time-series, and neural networks.

University of Arkansas

Teaching Assistant Jul 2012 - May 2017

A teaching assistant in the computer science department has a wide variety of courses that may be taught. This position involved teaching labs, teaching lectures, creating content, grading, holding office hours, and mentoring students. Taught over 400 students in many different course from Artificial Intelligence to Programming Foundations.

Lamar University & NSF

NSF REU Researcher May 2010 - Sep 2010

Researched large-scale databases for video, as well as methods for improving performance of mobile devices which resulted in a published paper.

Education

University of Arkansas

Aug 2012 - Dec 2015
Master's - Computer Science: Machine Learning GPA: 3.889

Northeastern State University

Jul 2008 - May 2012
Bachelor's - Computer Science GPA: 3.9

Publications

Practical Techniques For Using Neural Networks To Estimate State From Images
International Conference on Machine Learning Applications
Dec 2016

A novel way of estimating the state of a robot from camera images in a fast and efficeint manner.

Evaluating the Intrinsic Similarity between Neural Networks
University of Arkansas
Nov 2015

Thesis that examines methods for aligning and manipulating neural networks.

Modeling Time Series Data With Deep Fourier Neural Networks
Elsevier
Nov 2015

Creating deep fourier networks for time-series data that trained quickly and efficiently.

A Method for Finding Similarity between Multilayer Perceptrons by Forward Bipartite Alignment
International Joint Conference on Neural Networks
Jun 2015

A novel method for aligning neural networks, which enables ensembles and distributed training.

Training Deep Fourier Neural Networks to Fit Time-Series Data
Neuro and Evolutionary Computing
Apr 2014

Using the fourier transform to enable neural networks to better learn trends in time-series data.

IMISSAR: an intelligent, mobile middleware solution for secure automatic reconfiguration of applications, utilizing a feature model approach
ACM International Conference on Ubiquitos Information Management and Communication
Jan 2011

A middleware system that was designed to enable low-powered phones of the early 2010's to perform better by offloading intensive apps to desktop computers or servers.

Projects

Pancakes

A neural network library created in rust with the goal of being modular for neural network layers and blocks.

Robot Neural Decoder

A custom built neural network for estimating the complex state of a robot.

Dissertation

This Dissertation worked with robots and neural networks in an architecture that taught robot arms to perform tasks on their own.

Blockrunner

A winning project of the 2016 ACM/JBHunt Hackathon, this game featured mobile multiplayer assymetrical gameplay built with Jon Hammer.

UARK Social App

Winner of the 2014 ACM Hackathon, this mobile app used a multitude of different technologies including data mining, collaborative filters and more in an app that would help students at the University of Arkansas find walking instructions, recommend food on campus that they might like, and more.

Interests