Last week, Bouvet held its biannual internal conference-night called "Bouvet One". Here, we listened to each other speak about things ranging from concrete experiences during projects to cultural issues like how to increase diversity within our field.
I was bitten by the bug when I presented at our previous event half a year ago, after which the ball really got rolling for me with respect to presentations; and so I pitched three alternatives to our committee a couple of weeks ago - all of which were accepted.
Though it was an excellent situation to learn some new things, I had my work cut out for me - I had just under two weeks to do my research and finish the three presentations, which would have to be up to 20 minutes each. Luckily, I've talked quite a few times about our team's development methodology before, so I had much of what I needed in this case. I've also thought extensively about the various aspects of using webtech to make "native" applications before, and even submitted proposals for talks about this to several web- and JS-related conferences. Deepfakes, however, I didn't know much about; regrettably, I haven't found much time for Machine Learning projects since last Bouvet One, where I held a presentation about transfer learning - demoing a Star Trek Humanoid classifier I trained for a couple of minutes on my laptop's CPU, achieving an accuracy between about 79 and 95 percentage.
Since I'm posting a lot of slides these days, I've decided to add an IRL page I'll update with scheduled talks and appearances, slides and videos, press, etc.
All content was presented in Norwegian.
Go f**k yourself: How deepfakes ruined everything
A humourous, alarmist look at our changing relationship with the medium of video and with the truth, post deepfakes.
LEAN-machine: Goal-based development
Explanation of our development methodology - which consists of pieces from Scrum, KanBan, LEAN startup and DevOps - and how it lets us verify that are changes are for the good of the product.
Electron is bad and you should feel bad
A semi-humorous argument/rant about the downsides of using Electron and related technologies, including the technical and social implications of the resulting pseudo-native applications.