Moment-based Uniform Deviation Bounds for k-means and Friends
نویسندگان
چکیده
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 uniformly control deviations in the face of unbounded parameter sets, cost functions, and source distributions. To further demonstrate this mechanism, a soft clustering variant of k-means cost is also considered, namely the log likelihood of a Gaussian mixture, subject to the constraint that all covariance matrices have bounded spectrum. Lastly, a rate with refined constants is provided for k-means instances possessing some cluster structure.
منابع مشابه
Uniform 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 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 کاملSupersymmetry and the anomalous anomalous magnetic moment of the muon.
The recently reported measurement of the muon's anomalous magnetic moment differs from the standard model prediction by 2.6 sigma. We examine the implications of this discrepancy for supersymmetry. Deviations of the reported magnitude are generic in supersymmetric theories. Based on the new result, we derive model-independent upper bounds on the masses of observable supersymmetric particles. We...
متن کاملASSESSMENT OF DUCTILITY REDUCTION FACTOR FOR OPTIMUM SEISMIC DESIGNED STEEL MOMENT-RESISTING FRAMES
In the present study, ten steel-moment resisting frames (SMRFs) having different numbers of stories ranging from 3 to 20 stories and fundamental periods of vibration ranging from 0.3 to 3.0 second were optimized subjected to a set of earthquake ground motions using the concept of uniform damage distribution along the height of the structures. Based on the step-by-step optimization algorithm dev...
متن کاملOptimal convex combinations bounds of centrodial and harmonic means for logarithmic and identric means
We find the greatest values $alpha_{1} $ and $alpha_{2} $, and the least values $beta_{1} $ and $beta_{2} $ such that the inequalities $alpha_{1} C(a,b)+(1-alpha_{1} )H(a,b)
متن کامل