I just finished with my Skype a Scientist session! You can do it too if you want, just sign up. I had a lot of fun and while today was going to be on a slightly more depressing topic, I figure we can talk about how it went instead. I love doing outreach, it’s probably one of my favorite things about being a scientist and I got to show a whole bunch of interesting things I get to do as a scientist. This was an interesting group, so that’s also a good reason to talk about it.
Normally my outreach is to classrooms. I stick with older kids in high school getting ready for college because I like to mention that I didn’t have a direct pathway to education like some of them may find for themselves. I’ve given a class to kindergartners once, still not sure how trying to explain brain-machine interfaces to kids that young went, but I like to think I nailed it. At the very least they all seemed entertained!
Today I gave a talk to a small group, just four students and the instructor. It felt different talking to such a small group, but they picked me specifically so that was interesting. They were doing a project with prosthetics and had some very specific questions on machine learning, targeted muscle reinnervation, and they wanted to discuss a lot of the project they were working on. My sessions tend to last 30 to 45 minutes max, this ran a full hour.
I discussed with them my designs over the years, the things I’ve failed at, and how failure is okay too. They got to see some of the exoskeletons which I would normally demonstrate for them, but with COVID the demos come in the form of videos. I also got to go over some of the research I’m doing now and how it’s going to help people. Basically what I do as a researcher and why they may want to follow the same path.
Then we had the typical question and answer portion where I got to hear more about the work they are doing for class. They wanted my input on what they were doing and how machine learning may help them take the “next step” in the research they were reading about. I had used some machine learning in the past for another project so I went back to that and showed some of the results I got.
It was good because I applied an “easier” method first and then used machine learning. In one case the easier method was the better choice so it gave me a chance to talk about how you need to pick the right tool for the job. While I’m sure others may disagree, machine learning is like any other tool, there are some good times to use it and some bad times. Not every problem is best solved with a hammer, but a hammer can solve a lot of problems. That was the message I tried to convey to them, that while machine learning is a powerful tool that can do a lot of very interesting things, it may not give you the best outcome or there may be a comparably good outcome that is far easier to get.
In my case, I took the mean of a range of values and that predicted my two classes and got really good performance. Machine learning gave me slightly better, but was far more work and how it got the result was harder to interpret. In another case, the mean gave me horrible prediction. So I explained this to them in the hopes that they pick the right tool for the right job. It wasn’t to steer them away from machine learning, but more to point out that there are other tools that may work just as well and be much easier to use.
I really enjoy outreach because it gives me a chance to talk about the stuff I’m doing. I also end up answering a lot of questions you wouldn’t normally get from people in your field. Questions that make you think about things differently, or sometimes are just funny. Once I had a question about why the exoskeleton moved so slow and if we were going to make it go faster. Another person in that same group (from a different outreach event) asked why it was so loud.
These are obvious things, but sometimes answering them isn’t as straightforward. The exoskeleton moves slow because the person in it can’t walk so it can be jarring if it moved too quickly, but explaining that in a way that makes sense for someone younger isn’t super easy to do. The loud part, well that’s slightly easier to explain and yes, we’re trying to make them quieter, but that is a very exoskeleton specific thing. So needless to say some questions are hard, even the funny ones.
Well I think that’s it for today. Until next time! Oh and there’s still a pandemic going on, so be safe.