Featured Engineer

Interview with Deniz Erdogmus

Deniz Erdogmus

Deniz Erdogmus - Assistant Prof, ECE, Northeastern University

How did you get into electronics/ engineering and when did you start?

In high school I thought I would study physics and mathematics to pursue research in theoretical physics, specifically on the theory of everything. When the time to apply to departments at the end of high school, I ended up selecting electrical engineering as a good compromise considering career options, expected income, and similar factors. In Turkey, where I received my BS degree, one chooses the desired departments and universities during a centralized placement exam. I performed well enough to get into my first choice. I got a double major in math, though, to satisfy my high school plans partly.

What are your favorite hardware tools that you use?

I am primarily a math oriented person and I enjoy algorithm design and theoretical analysis using statistical methods more than anything. However my group performs research on brain computer interfaces (BCI) and biomedical image processing. For BCI we have to use bioamplifiers to acquire EEG and develop a real-time system.

What are your favorite software tools that you use?

My favorite is Matlab, although it is too slow for most heavy-duty computations that my students do in the lab. I design basic proof-of-concept algorithms for illustration, so Matlab is sufficient fo my personal purposes. The students, then struggle with making it faster and capable of handling larger amounts of data using C and C++. I look forward to reliable and easy GPU computing in Matlab – I did not have the chance to try the various options for this purpose yet, but I have not heard from colleagues that one of these is really useful and superior either.

What is the hardest/trickiest bug you have ever fixed?

The gradient descent optimization algorithm for a multilayer recurrent neural network training algorithm implementation was the trickiest code I had to make work. It ended up being very slow and I followed up by designing an Extended Kalman Filter based optimization algorithm which was much simpler and faster to converge.

What is on your bookshelf?

Various textbooks on signal processing and machine learning. Complimentary copies of journal articles and books that I published, if the publisher had sent me one. Dissertations that are written by me, my students, and others that I find interesting to keep close-by.

Do you have any tricks up your sleeve?

When writing code, I make sure that I have all equations on paper with a good notation, then ensure that the variable names and function calls in the code use the same notation so that the code reads like my equations. This helps me understand what my code is doing in terms of equations even if I read it 1 year later.

What has been your favorite project?

The real-time brain computer interface typing interface we are developing with colleagues at OHSU is my favorite project so far. Our locked-in subjects who try the system and give us critical feedback for making improvements to suit their needs better also provide us very satisfying positive feedback. One such subject said that he felt his head perspire after each correct letter selection using our brain interface prototype in a follow up email he sent using his usual head-mouse system. Seeing that we are developing a product that will be appreciated by the users and that will improve the quality of life for them is great.

Do you have any note-worthy engineering experiences?

Developing various assistive brain interface technologies with undergraduate students at Northeastern as part of the capstone project requirement had been one of the most satisfying accomplishments for me. Working with seniors who take an idea and make a working prototype over a couple of months, all the struggles to get over obstacles during the process, and experiencing the final moment of demonstrating a successfully working product at the end of the academic year is a great experience. Seeing the professional growth of the students from excited and interested students to capable and confident engineers during the process is priceless. Projects we have designed so far can be viewed at my website.

What are you currently working on?
  1. Designing brain interfaces to communicate with computers and to control robots and devices.
  2. Developing new algorithms for automatic image analysis and understanding to speed up scientific and medical discovery.
What direction do you see your business heading in the next few years?

I will be continuing to work on the two application areas of signal and image processing: brain interfaces, biomedical image analysis. I hope to have a variety of systems and software packages that will impact the lives and research of other people.

What challenges do you foresee in our industry?

In my opinion, the lack of mathematical rigor and the desire for quick turn-around to meet sales-oriented deadlines causes poor products to be delivered to the market. These products could be things we purchase or they could be papers written. The availability of sheer computational power tends to make students in my area to forget that number crunching does not solve problems, intelligently and carefully developed mathematical constructs do; the number crunching algorithms are merely a specific implementation of these mathematical ideas. I hope that students are pointed out to this fact and they think first, implement later. This probably goes for all kinds of design processes.

Previous Spotlights

x
Like free stuff?
EEWeb Weekly Giveaway Sponsored by Mouser This Week: TI Low Profile NEXFet power supply dev kit!
Enter Here
Login and enter if you're already a member.
Click Here