ckrapu

13 ms·

1. 181.▲
   by ckrapu 4y ago

   I looked over the author’s 2018 paper which is linked in the press release and I think he pretty squarely addresses the issues you raise. In particular, check out figure 1B in that work for one particularly clear signal of change over time
2. 182.▲
   by ckrapu 4y ago

   What do you think comes next after HMC/NUTS for general purpose turn-key sampling?
3. 183.▲
   by ckrapu 4y ago

   To play along more with the ideas you guys are raising, I think that strong spatial heterogeneity driven by falling birth rates in sink-type effects are having an outsized effect on how we think about nations even if the sink is due to dens
4. 184.▲
   by ckrapu 4y ago

   My sense was that this interpretation of work on the “behavioral sink” is quite off. I think most reasonable biologists and ecologists would agree that density-dependent stress tends to cause the pathologies of the type that are described i
5. 185.▲
   by ckrapu 5y ago

   The heuristic I learned in a Bayesian filtering class is to use a diagonal matrix and inflate the prior predictive variance by a small amount each iteration (1-5%) so that the data can easily tamp it down if there’s any kind of signal. The
6. 186.▲
   by ckrapu 5y ago

   They are better than most VAE models at not averaging, but they may still not be covering the entire data distribution. That would probably imply that the underdispersed samples are closer to the average. We don’t have great metrics for fig