S-based Inference III

Now: x from unknown suboptimal process on unknown machine

Postulate still true: resulting S’(x) ə/t if x costs > O(t)

Then dominance: S(x) > cS’(x) for all x

We can still use AS to predict well! According to Hutter’s loss bound:

Previous slide Next slide Back to first slide View graphic version

Back to J. Schmidhuber's Speed Prior page