Portland State University Computer Science Department Maseeh College of Engineering and Computer Science

Mick Thomure
Doctoral Student

Contact Information

Computer Science Department
Portland State University
P.O. Box 751
Portland, OR 97201

Email: thomure@cs.pdx.edu
Phone: (503) 725-4059


In Progress: PhD in Computer Science, Portland State University, Portland, OR.

BS in Computer Science, Spring 2006, Portland State University, Portland, OR.

Research Interests

I am interested in how we organize data, and how this is changing. I believe the responsibility of filtering information is shifting from those generating the data to those consuming it. This change is a good thing, allowing data to be used in novel and unanticipated ways, but it also means data consumers require new tools to adapt. I aim to invent these tools; creating massively-scalable software systems that use statistical modeling to provide on-demand, user-centric filtering and organization of data.

My research interests include the fields of machine learning, in which we use statistical models to capture patterns in data; information retrieval, which allows us to filter and organize large amounts of data in an automated and scalable way; and computer vision, which allows us to apply the above topics to the challenges presented by visual data.


My dissertation focuses on improving a class of biologically-inspired object recognition systems. This is an exciting area that brings together computer vision, machine learning, and computational neuroscience. The goal of my work is to increase the usefulness of these systems as practical tools for computer vision, while simultaneously expanding our understanding of them as neuroscience models.

Part of this work has been the development of Glimpse, an open-source library for implementing hierarchical visual models in C++ and Python.

More Info | Glimpse Project


Michael D. Thomure, Melanie Mitchell, Garrett T. Kenyon (2013). On the Role of Shape Prototypes in Hierarchical Models of Vision. To appear in The International Joint Conference on Neural Networks (IJCNN).

Will Landecker, Michael D. Thomure, Luis M. A. Bettencourt, Melanie Mitchell, Garrett T. Kenyon, Steven P. Brumby (2013). Interpreting Individual Classifications of Hierarchical Networks. To appear in Computational Intelligence and Data Mining - CIDM 2013, Special session on Interpretable Systems in Machine Learning.

Michael Thomure, Will Landecker and Melanie Mitchell (2011). Random prototypes in hierarchical models of vision. In Grand Challenges in Neural Computation II (poster). PDF

Will Landecker, Michael Thomure and Melanie Mitchell (2011). Background cues in images classified by hierarchical models. In Grand Challenges in Neural Computation II (poster).

Will Landecker, Steven P. Brumby, Mick Thomure, Garrett T. Kenyon, Luis M. A. Bettencourt and Melanie Mitchell (2010). Visualizing classifications of hierarchical models of cortex. In Computational and Systems Neuroscience - COSYNE 2010 (poster).

Melanie Mitchell, Michael D. Thomure, and Nathan L. Williams (2006). The role of space in the success of coevolutionary learning. In Proceedings of Artificial Life X: Tenth Annual Conference on the Simulation and Synthesis of Living Systems. PDF

Bart Massey, Mick Thomure, Raya Budrevich, and Scott Long (2003). Learning spam: Simple techniques for freely-available software. In USENIX Annual Technical Conference, FREENIX Track. PDF


Perception and Analogy-Making in Visual Understanding. Spring '06 - Present
The purpose of this work is to extend Mitchell and Hofstadter's Copycat system to the problem of image interpretation. In the Spring of 2008, I presented some results during my research proficiency exam.
More Info | Professor Mitchell Homepage

Coevolution: Studying the effectiveness and dynamics of coevolutionary learning. Fall '04 - Summer '06
The purpose of this work is to better understand the characteristics of this adaptive algorithm which distinguish it from related techniques such as resource sharing. In particular, the effect of using spatial orientation on the speed and efficacy of searching a hypothesis space is being explored. This work is being performed as an undergraduate RA for Professor Melanie Mitchell.
More Info | Professor Mitchell's Homepage

MozTorch: Bringing machine learning to the Mozilla community. Spring '05 - Summer '05
This project was funded by Google through their Summer of Code program.
Project Homepage | SoC Homepage

Sensnet: The PSU Sensor Network Project. Spring '04 - Summer '04
This work was performed as an undergraduate research project with the goal of developing remote sensing applications using networks of wireless motes. The project mentors where Professor Suresh Singh and his graduate research assistant Julie Lee.
Project Homepage | Professor Singh's Homepage

PSAM: Applying machine learning to spam recognition. Fall '01 - Fall '03
The goal if this project has been to contribute machine learning code and benchmark corpora to the open-source community, in particular those working on spam filtering. The project has developed a C library of useful machine learning algorithms, and presented an empirical comparison of those algorithms at the USENIX Conference.
Project Homepage | USENIX Paper | Professor Massey's Homepage

PSU Intelligent Robotics Lab: Electric Horse Project. Fall '00 - Spring '01
Participated as mentor in a project that brought high school students into the Intelligent Robotics Laboratory at Portland State. The goal of this project was to teach students about core electrical and computer engineering concepts through the development of a robotic horse that can walk in the Rose Parade.
Project Homepage | Professor Perkowski's Homepage