Brad Wyble    @bradpwyble    11/23/2021      

The Meta-Bayes theorem dramatically simplifies the math.
  
    1         13










 
  Related  

Andrej Karpathy    @karpathy    11/30/2021      

2/3 The multicam + video data, temporal continuity of a slowly moving viewpoint, close collaboration with data sourcing and labeling, and the infinity-sized dataset of unlabeled clips dramatically expands creative modeling opportunities on the neural net side
  
    40         519




Computationally sound arguments for 3-coloring via PCP theorem and Merkle trees: https://t.co/rOYFxRKqyH
  
    1         1



Alex Dimakis    @AlexGDimakis    11/6/2021      

One more thing on Shannon-Nyquist: the fact that you can roughly approximate any smooth function by sampling and interpolating is obvious. The theorem is that you can *perfectly reconstruct* a continuous function if its band limited and you interpolate with Sinc.
  
    1         26



Alex Dimakis    @AlexGDimakis    11/6/2021      

Here is a very good reason why the Nyquist–Shannon sampling theorem requires that your function is low-pass before you sub-sample to downscale. If you just sub-sample without smoothing, a bad guy can place another image exactly on the pixels you sub-sample. Adversarial aliasing.
  
    59         376