Large deviation bounds for k -designs
نویسندگان
چکیده
منابع مشابه
Large Deviation Bounds for k-designs
We present a technique for derandomising large deviation bounds of functions on the unitary group. We replace the Haar distribution with a pseudo-random distribution, a k-design. kdesigns have the first k moments equal to those of the Haar distribution. The advantage of this is that (approximate) k-designs can be implemented efficiently, whereas Haar random unitaries cannot. We find large devia...
متن کاملUniform Deviation Bounds for k-Means Clustering
Uniform deviation bounds limit the difference between a model’s expected loss and its loss on a random sample uniformly for all models in a learning problem. In this paper, we provide a novel framework to obtain uniform deviation bounds for unbounded loss functions. As a result, we obtain competitive uniform deviation bounds for k-Means clustering under weak assumptions on the underlying distri...
متن کاملLarge Deviation Bounds for Markov Chains
We study the fraction of time that a Markov chain spends in a given subset of states. We give an exponential bound on the probability that it exceeds its expectation by a constant factor. Our bound depends on the mixing properties of the chain, and is asymptotically optimal for a certain class of Markov chains. It beats the best previously known results in this direction. We present an applicat...
متن کاملUniform Deviation Bounds for Unbounded Loss Functions like k-Means
Uniform deviation bounds limit the difference between a model’s expected loss and its loss on an empirical sample uniformly for all models in a learning problem. As such, they are a critical component to empirical risk minimization. In this paper, we provide a novel framework to obtain uniform deviation bounds for loss functions which are unbounded. In our main application, this allows us to ob...
متن کاملMoment-based Uniform Deviation Bounds for k-means and Friends
Suppose k centers are fit to m points by heuristically minimizing the k-means cost; what is the corresponding fit over the source distribution? This question is resolved here for distributions with p ≥ 4 bounded moments; in particular, the difference between the sample cost and distribution cost decays with m and p as mmin{−1/4,−1/2+2/p}. The essential technical contribution is a mechanism to u...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
ژورنال
عنوان ژورنال: Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences
سال: 2009
ISSN: 1364-5021,1471-2946
DOI: 10.1098/rspa.2009.0232