Being a CS major, I would have loved to have heard more about the technical details of the deep learning that Meade was doing to predict earthquakes. There was a discussion of how little data there is, and I know that's always a very difficult issue to overcome. Furthermore, what inputs and outputs one should use are very non-obvious.
top of page
bottom of page
While the lack of technical content made sense for the audience, I would love to know more about the details. While weather predictions can take into account satellite imagery, real time weather data from on the ground locations, and a number of other always-recording sensors, data is much less plentiful for earthquake prediction. While you can set up seismic monitoring stations, I'd imagine these are expensive, and only give data very infrequently (i.e. when small earthquakes are felt).