Speaker Spotlight: Dr. Michael Gashler
A developer conference in Northwest Arkansas would not be complete without one of the brightest minds currently working for the University of Arkansas.
Introducing one of our featured speakers: Dr. Michael Gashler, Professor of Computer Science and Computer Engineering at the University of Arkansas!
Bio: Michael S. Gashler studies in the field of machine learning. His research interests focus on methods that enable machines to perceive and understand high-dimensional observations, and to operate effectively in circumstances that were not anticipated a priori by the programmer. Dr. Gashler earned his Ph.D. in machine learning with a focus in non-linear dimensionality reduction from Brigham Young University in 2012. Prior to his graduate studies, he worked for Microsoft where he helped to develop the security system underlying the Common Language Runtime. He is the architect of several open source projects, including "Waffles", a toolkit of machine learning and data mining algorithms. He taught himself to program when he was young by trying to build the video games that his parents did not give to him. He now hopes to inspire students to build technology that science has not yet given to us. In particular, his lab intents to narrow the gap between the cognitive capabilities of humans and machines.
Here’s the 411 on Dr. Gashler’s talk about neural networks.
Topic: Using Neural Networks to Solve Non-Traditional Challenges
The Breakdown: It is well understood that deep neural networks can be highly effective function approximators, but it is less well known that function approximation is not the only thing they can do. Neural networks can also infer missing values in data, generate anticipated observations, estimate state in dynamic systems, remove noise from inputs, reduce dimensionality, and fuse information from different modalities. Applying neural networks in such atypical ways requires the user to have a little knowledge about how they work internally. This talk seeks to impart a little intuition for what happens inside a neural network, and will demonstrate how neural networks can be used in applications that require a more than just function approximation. The goal of this talk will be to enable the attendees to make more effective use of neural networks in non-traditional applications.
Dr. Gashler answered a few fun questions for us.
How did you get started with programming?
When I was 12, my friends all had game consoles. I begged for one, but my dad bought a programmable computer instead. I spent years trying to recreate the games I wanted. By the time I was good enough to actually do it, I found that I enjoyed making more than playing.
What drew you to the University of Arkansas?
When I completed my PhD, I applied at 64 universities. Eight of them responded for a phone interview. Four of these flew me out for an on-site interview. Two of those made me an offer. The people at U of A seemed less stressed, so I accepted, and here I am.
What does your day-to-day at work look like?
Being a professor involves more multitasking than most people realize. I teach only two classes this semester, but I have well over 100 students. In any given half-hour, at least one of them needs me to drop whatever I’m trying to think about and focus on his or her problem for a while. By the end of the day, I have a vague recollection of having helped a lot of people, and I find great satisfaction in that, but I really couldn’t account for what I did during a day if my life depended on it. A few times each year a deadline for a grant proposal approaches. I am perpetually disappointed at myself for waiting until just before the deadline to get started. I squeeze in the needed time by skipping lunch and sleep. My inbox rarely has fewer than 30 unanswered emails hanging over my head. But, every few days, one of my students who was previously struggling announces that he or she “gets it”, or one of my graduate students takes an idea I shared with him and makes more out of it than I even foresaw, and I realize that no other job could ever make me feel as satisfied.
What is your favorite aspect of Northwest Arkansas?
The people here are warm and genuine.
Mac, PC or Linux?
Linux. Live free or die.
Who is your favorite character on “Silicon Valley”?
Is that a TV show? I haven’t watched one of those in about seven years.
What is a fact people would be surprised to know about you?
I made my first Facebook account (under duress) last month.
What is your least favorite language?
English? It’s so nuanced, ambiguous, and imprecise. I’m not a fan of Perl either.
Musk or Bezos?
I appreciate OpenAI, so Musk.
Bitcoin. I tried and failed to solve the double-spending problem years earlier for a virtual economy in a decentralized MMO I was trying to build. When Satoshi Nakamoto succeeded where I had failed, I knew I had to buy in. Alas, I moved most of my bitcoins into MtGox at a most unfortunate time.