AI-powered
podcast player
Listen to all your favourite podcasts with AI-powered features
Avalanche Images - Do You Find the Threshold Equal?
Did you run any xperiments to try to understand the kind of nuance there? Not directly. We did things like drill down and test only on even less balanced data sets, like avalanche carters in the alps that were known to have a lot of avalanches. And that kind of actually increased the lift among the methods. But it would be interesting to try to derive. So, i mean, another thing my group is really interested in is interpretability. The other paper i was going to talk about was about interpolations through latent space to bring two domains closer together, called domain alignment. You bring up a really interesting interpretability question that we c maybe look at in future work