Пікірлер

• 
  @matej64189 ай бұрын

  How does online inference step work then? Say I observe additional accidents in coal or airlines, but there is no time to retrain the GP. Do I simply add observations to GP data and then predict for unseen steps?

• 
  @dilipsenapati84142 жыл бұрын

  Nice Lecture

• 
  @karangurtu2 жыл бұрын

  Thanks Professor Solin for crisp and concise explanations!

• 
  @calebjuma76483 жыл бұрын

  very insightful. What if the stroller is not stopped intermittently, but rather only at the starting and destination point...Will the biases be too big to create big positional errors?

• 
  3 жыл бұрын

  Nice video, could you share code? I am interesting in tiny trajectories but I am not sure what error to expect.

• 
  @nguyenngocly14843 жыл бұрын

  You can have swapped around neural nets too. With fixed dot products (enacted with fast transforms) and adjustable activation functions. Parametric (adjustable) ReLU is anyway a known thing. Of course you have to prevent the first transform from taking the spectrum of the input which you can do using a random fixed pattern of sign flips. And use a final transform as a readout layer. The fast Walsh Hadamard transform is good.