J. Schmidhuber. The Speed Prior: a new simplicity measure yielding near-optimal computable predictions. In J. Kivinen and R. H. Sloan, editors, Proceedings of the 15th Annual Conference on Computational Learning Theory (COLT 2002), Lecture Notes in Arti cial Intelligence, pages 216-228. Springer, Sydney, Australia, 2002.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....idea has been used in [LW94, Vov92] for sequence prediction, and is referred to as weighted majority there. 6. 1 Time Limited Probability Distributions In the literature one can find time limited versions of Kolmogorov complexity [Dal73, Dal77, Ko86] and the time limited universal semimeasure [LV91, LV97, Sch02b]. In the following, we utilize and adapt the latter and see how far we get. One way to define a timelimited universal chronological semimeasure is as a sum over all enumerable chronological semimeasures with computation time at most t and of size at most l. yx 1:n ) # : l(#)# l ....

.... and Warmuth [LW94] universal forecasting by Vovk [Vov92] Levin search [Lev73] pac learning introduced by Valiant [Val84] and Minimum Description Length [LV92a, Ris89] Resource bounded complexity is discussed in [Dal73, Dal77, FMG92, Ko86, PF97] resource bounded universal probability in [LV91, LV97, Sch02b]. Implementations are rare and mainly due to Schmidhuber [Con97, Sch97, SZW97, Sch02a] Excellent reviews with a philosophical touch are [LV92b, Sol97] For an older, but general review of inductive inference see Angluin [AS83] Sequential decision theory. The other ingredient in our AI# model is ....

J. Schmidhuber. The Speed Prior: a new simplicity measure yielding near-optimal computable predictions. In J. Kivinen and R. H. Sloan, editors, Proceedings of the 15th Annual Conference on Computational Learning Theory (COLT 2002), Lecture Notes in Artificial Intelligence, pages 216--228, Sydney, Australia, July 2002. Springer.

No context found.

J. Schmidhuber. The Speed Prior: a new simplicity measure yielding near-optimal computable predictions. In J. Kivinen and R. H. Sloan, editors, Proceedings of the 15th Annual Conference on Computational Learning Theory (COLT 2002), Lecture Notes in Arti cial Intelligence, pages 216-228. Springer, Sydney, Australia, 2002.

No context found.

J. Schmidhuber. The Speed Prior: a new simplicity measure yielding near-optimal computable predictions. In J. Kivinen and R. H. Sloan, editors, Proceedings of the 15th Annual Conference on Computational Learning Theory (COLT 2002.

No context found.

J. Schmidhuber. The Speed Prior: a new simplicity measure yielding near-optimal computable predictions. In J. Kivinen and R. H. Sloan, editors, Proceedings of the 15th Annual Conference on Computational Learning Theory (COLT 2002.

No context found.

J. Schmidhuber. The Speed Prior: a new simplicity measure yielding near-optimal computable predictions. In J. Kivinen and R. H. Sloan, editors, Proceedings of the 15th Annual Conference on Computational Learning Theory (COLT 2002), Lecture Notes in Arti cial Intelligence, pages 216-228. Springer, Sydney, Australia, 2002.

....typical ecient MDL approaches require the speci cation of a class of computable models of the data, say, certain types of neural networks, plus some computable loss function expressing the coding costs of the data relative to the model. This provokes numerous ad hoc choices. Our recent work [55], however, o ers an alternative to the celebrated but noncomputable algorithmic simplicity measure or Solomono Levin measure discussed above [59, 75, 60] We introduced a new measure (a prior on the com putable objects) which is not based on the shortest but on the fastest way of describing ....

....following postulate. Postulate 1 The cumulative prior probability measure of all x incomputable within time t by any method is at most inversely proportional to t. This postulate leads to the Speed Prior S(x) the probability that the output of the following probabilistic algorithm starts with x [55]: 1. Toss an unbiased coin until heads is up; let i denote the number of required trials; set t : 2 . 2. If the number of steps executed so far exceeds t then exit. Execute one step; if this leads to a request for a new input bit (of the growing self delimiting program, e.g. 30, 31] ....

[Article contains additional citation context not shown here]

J. Schmidhuber. The Speed Prior: a new simplicity measure yielding near-optimal computable predictions. In J. Kivinen and R. H. Sloan, editors, Proceedings of the 15th Annual Conference on Computational Learning Theory (COLT 2002), Lecture Notes in Arti cial Intelligence, pages 216-228. Springer, Sydney, Australia, 2002.

.... speed ups due to halting programs if there are any) Nonbinary, nonuniversal variants of Osearch were used to solve machine learning toy problems unsolvable by traditional methods [58, 47] Probabilistic alternatives based on probabilistically chosen maximal program runtimes in Speed Prior style [41, 45] also outperformed traditional methods on certain toy problems [39, 40] 2.4 Incremental Search Since Newell Simon s early attempts at building a General Problem Solver [32, 35] much work has been done to develop mostly heuristic machine learning algorithms that solve new problems based on ....

....most recent code and prolongations thereof. Yet other oops variants will also assign fractions of the total time to the second most recent program and its prolongations, the third most recent program and its prolongations, etc. We may also consider probabilistic oops variants in Speed Prior style [41, 45]. One not necessarily useful idea: Suppose the number of tasks to be solved by a single program is known in advance. Now we might think of an OOPS variant that works on all tasks in parallel, again spending half the search time on programs starting at a last , half on programs starting at a ....

[Article contains additional citation context not shown here]

J. Schmidhuber. The Speed Prior: a new simplicity measure yielding near-optimal computable predictions. In J. Kivinen and R. H. Sloan, editors, Proceedings of the 15th Annual Conference on Computational Learning Theory (COLT 2002.

No context found.

J. Schmidhuber. The Speed Prior: a new simplicity measure yielding nearoptimal computable predictions. In Proceedings of the 15th Annual Conference on Computational Learning Theory (COLT 2002), Lecture Notes in Artificial Intelligence, pages 216--228, Sydney, Australia, 2002. Springer.

No context found.

J. Schmidhuber. The Speed Prior: a new simplicity measure yielding near-optimal computable predictions. Proceedings of the 15th Annual Conference on Computational Learning Theory (COLT 2002), 2002.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC