. @Ylecun are you arguing that your team has a robust solution to the problem of getting deep nets to understand the causal consequences of events as they unfold over time, or just pointing me to a toy model? Have you tried it on the examples in this thread?https://twitter.com/ylecun/status/1188902027495006208 …
-
It’s great to keep everybody in the community accountable to actual results and actual performance. It’s also highly necessary to also keep hype way down. At the same time, we shouldn’t minimize incremental results.
that paper is interesting; i just don't think that it is reasonably characterized as rendering my claim false, let alone a "rearguard action". a fairer claim would be "there's some work on that in a limited domain, which I find promising." I'd be fine with that.
