A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2019; you can also visit the original URL.
The file type is
Our approach generalizes both UCB and LinUCB to far more expressive possible model classes and achieves low regret under certain distributional assumptions. ... Our algorithms leverage the availability of a regression oracle for the value-function class, a more realistic and reasonable oracle than the classification oracles over policies typically assumed by agnostic ... Note that RegCB is not minimax optimal: while it can obtain O KT log F regret or even logarithmic regret under certain distributional assumptions, which we describe shortly, for some instances it can make ...arXiv:1803.01088v1 fatcat:f2nmxg3izvbifhr3m2fg72uftu
Lecture Notes in Computer Science
dimensional distributions Regular vines are a graphical tool for representing complex high dimensional distributions as bivariate and conditional bivariate distributions. ... for high-dimensional classification. ... Most studies, however, are rather focused on the cases with the single dimensional (or at least low dimensional) function parameter in the regression model since the dimensionality costs higher order smoothness ...doi:10.1007/11551188_19 fatcat:gtzi5u6emvfyxea3zbklobnvom
Often times, these models result in stochastic programs with high-dimensional recourse vectors. ... In the application section, the proposed approach is compared with traditional neural networks in terms of their performance against the complex problems Anton J. ... This prospective is used for example in deciding which capacity we need to expand over the next years. ...fatcat:b2q6nnj52nbudkna2lpsjmhtlq