Open Research Positions
If you are from outside Clark University, please do not send cold email inquiries to join my research lab. Due to volume of messages, I am unable to respond.
I am always looking for undergraduate and graduate researchers to build out basic demonstrators into more functional prototypes and contribute to research projects. I generally do not take on a research student unless they have taken a course with me or come highly recommended by a colleague. This helps me to asess your suitability for my research lab. Moreover, for the undergraduate student researcher, requirements include a requisite mathematical background (ideally Calculus, Discrete Math, Stochastic Computing) and ability to design and write software in a structured programming language (C++, Java, Python). You must understand data structures. It is also important that you enjoy tinkering, have an excellent work ethic, and are comfortable living in the space of the unknown where concepts and materials are not found in a textbook. For the graduate student researcher, requirements include enjoyment of tinkering and reduction to practice by implementation of mathematical models or algorithmic concepts. Additionally, the graduate student must either have taken a Machine Learning course or have the equivalent knowledge. My research spans the very theoretical through applied and typically involves sigificant engineering work with sensors, mechanisms, and systems. Depending on available budgets, I try to fund research positions. If funding is not available, independent study credits can be arranged for this work. Because of the significant investment in training involved, preference in consideration is given to longer term (2+ years) commitments to research. Please do not assume your interest is a guarantee to work in my lab. I reserve the right to turn down requests to become involved in research in my laboratory.
I also consider Senior Capstone students. My preference is for 1-semester Capstone research projects. I always consider graduate thesis students. If you are an undergraduate or graduate
student at Clark University and are interested, please contact me by via your Clark email only to arrange an appointment.
My research students tend to have an excellent record of success including research publications,
numerous awards, graduate school admissions, and industry positions. My outcomes include a 100% placement
rate among my research students.
About Gary
Gary Holness is an associate professor in the Department of Computer Science at Clark University (ClarkU).
At Clark he directs the Laboratory for Intelligent Perceptual Systems (LIPS). He has been at Clark University since Fall 2021. Over his career, he has been a PI or co-PI on competitive grants totalling over $5.3M dollars. At Clark University he has directed 6 research students (5 undergrad, 1 M.S.) and 1 significant semester-long Senior Capstone project.
Prior to Clark, he spent 11 years at Delaware State University (DSU). At DSU, he played a significant role in designing and launching a new graduate program, has contributed to the complete re-design of two undergraduate curricula (BS in Computer Science, BS in Information Technology), and has advised 44 research students ( 2 PhD, 11 MS, 31 BS). He launched a new MS program in Computer Science and served as the inaugural Graduate Program Director. He was key in the planning and construction of a significant maker-space facility at DSU that includes capabilities for plastic and metal 3D Printing, PCB printing, 3D scanning, soldering and electronics fabrication, etching and cutting, polishing and finishing, caprentry, and 5 Axis CNC milling. As a principle in DSU's maker space effort, he created the vision, developed the budget, specified and lead the acquisition of equipment and supplies. Prior to joining Delaware State University, he was a Lead Research Scientist in the Artificial Intelligence and Brain Inspired Computing research groups at Lockheed Martin Advanced Technology Laboratories.
Gary received his Ph.D. in computer science from the University of Massachusetts, Amherst (UMass),
focusing on machine learning, robotics, distributed systems, machine perception, and statistics. His research interests include Machine Learning, Robotics, Statistics, Distributed Systems, and real-word Machine Learning applications. His PhD thesis contributed a new method for machine learning ensembles. He developed algorithms for training ensembles in a control theoretic framework that exercises direct control over component error distributions by sharing informatin about traning set selection. His algorithm, Discrete Instance Selection and Collective Optimization (DiSCO), guides the construction of ensembles whose component classifiers select complementary error distributions, sometimes making suboptimal individual choices that maximize diversity while minimizing overall ensemble error. Treating ensemble construction as an optimization problem, he developed approaches using local search, global search, and Jaynes' MaxEntropy framework.
At UMass, he was a member of the Machine Learning Lab where he was advised by Professor Paul Utgoff (memorial). His thesis work was supported by the National Science Foundation (award number ATM-0325167) on a collaboration with the UMass Computer Vision Laboratory and Bigelow Laboratories for Ocean Science. While at UMass, he also spent many fruitful years working in the Laboratory for Perceptual Robotics on distributed robot teams and Smart Room research under the advisement of Professor Rod Grupen.
His collaborators (past and present) have included Sokratis Makrogiannis, JinJie Liu, Hacene Boukari, Gour Pati, Tomasz Smolinski, Dragoljub Pokrajack, Allen Hanson, Rod Grupen, Howard Schultz, Dimitri Lisin, Marwan Mattar, Sai Ravela, Matthew Blasckho, Michael Seiracki, Eric Eaton,
Dan McFarlane (LM Fellow, Lockheed Martin Advanced Technology Laboratories); In Memoriam: Paul Utgoff, Ed Riseman.
Computation should be an active part of the world, situated within it, gathering observations and acting
upon them. In support of this vision, my research rests at the seams connecting machine learning,
Robotics, Statistics, Distributed Systems, and cyber-physical systems. I am also excited about real world
applications of machine learning. Broad topic areas of interest include:
- Ensemble methods: controlling error diversity and understanding ensemble dynamics
- Information Theoretic Methods: novel methods for measurement and charaterization of regularities in data
- Methods for density estimation: parametric, non-parametric, semi-parametric
- Representation learning: methods that learn features of the environment to best support learning
- Intelligent Sensing:pattern discovery from sensor data, synthesis across multiple modalities
- Machine Perception: methods for long-term sparse representations enabling systems to discern and reason over the content of their environment from a variety of sensory modes
- Cyber-Physical Systems: reactive environments that marry physical processes with computational processes, sensing, and communication to respond quickly to the non-stationary dynamics of everyday things
- Internet of Things: networked physical devices, vechicles, buildings, etc. embedded with electronics, software, sensors, and actuators.
- Clinical Informatics: sensing, pattern discovery, integration, reasoning, and dissemination of indicators for medical events from clinical data streams.
- Distributed Systems: frameworks, software, methods
- Systems Integration: rapid prototyping
- Research Training: novel methods for training and mentoring graduate and undergraduate students
Details of my research on these topics can be found on my research and publications pages. This research has also produced a number of software packages, which I make freely available for academic and not-for-profit use.
Contact Information
|
Mailing Address: |
E-mail:
|