Filters








627,761 Hits in 7.9 sec

Being Robust (in High Dimensions) Can Be Practical [article]

Ilias Diakonikolas, Gautam Kamath, Daniel M. Kane, Jerry Li, Ankur Moitra, Alistair Stewart
2018 arXiv   pre-print
Robust estimation is much more challenging in high dimensions than it is in one dimension: Most techniques either lead to intractable optimization problems or estimators that can tolerate only a tiny fraction  ...  Recent work in theoretical computer science has shown that, in appropriate distributional models, it is possible to robustly estimate the mean and covariance with polynomial time algorithms that can tolerate  ...  Our Results Our goal in this work is to show that high-dimensional robust estimation can be highly practical. However, there are two major obstacles to achieving this.  ... 
arXiv:1703.00893v4 fatcat:j5xbd6knhzgyhifkl5pqmbdd6y

Outlier identification in high dimensions

Peter Filzmoser, Ricardo Maronna, Mark Werner
2008 Computational Statistics & Data Analysis  
A computationally fast procedure for identifying outliers is presented, that is particularly effective in high dimensions.  ...  The performance of this method is illustrated on real and simulated data with dimension ranging in the thousands.  ...  Practically, we found good results using a level of 99%. It can be argued this produces similar results to transforming the data via SVD to a dimension less than the minimum of n and p.  ... 
doi:10.1016/j.csda.2007.05.018 fatcat:gkdwonkx5ffgvi4iqwloh7o4sy

Statistical Optimization in High Dimensions

Huan Xu, Constantine Caramanis, Shie Mannor
2016 Operations Research  
In the important case where noise artificially increases the dimensionality of the parameters, we show that combining robust optimization and dimensionality reduction can result in high-quality solutions  ...  This setup falls far outside the traditional scope of Robust and Stochastic optimization.  ...  (This assumption is needed to facilitate the analysis but can be ignored in practice by projecting the set to which the solution belongs to.)  ... 
doi:10.1287/opre.2016.1504 fatcat:3pwzncsxjvgytozjl6ytuxmdae

Global Development and Happiness: How Can Data on Subjective Well-Being Inform Development Theory and Practice?

Christian Kroll
2013 IDS Working Papers  
Summary How can the new science of happiness add value to development theory and practice?  ...  order to transform thinking, policy and practice.  ...  the implications for development practice could be.  ... 
doi:10.1111/j.2040-0209.2013.00432.x fatcat:tmbr2u77qff4png6dkztzwz6qu

Averaging causal estimators in high dimensions [article]

Joseph Antonelli, Matthew Cefalu
2020 arXiv   pre-print
Averaging is widely used in statistics for problems such as prediction, where there are many possible models, and averaging can improve performance and increase robustness to using incorrect models.  ...  We show that these ideas carry over into the estimation of causal effects in high-dimensional scenarios.  ...  We show both theoretically and empirically that this leads to robust performance for estimating treatment effects in high-dimensions.  ... 
arXiv:1906.09303v2 fatcat:mes2wgyxu5hprok5flwqnev3me

Profile Analysis in High Dimensions

Cigdem Cengiz, Dietrich von Rosen, Martin Singull
2020 Journal of Statistical Theory and Practice  
AbstractThe three tests in profile analysis: test of parallelism, test of level and test of flatness are modified so that high-dimensional data can be analysed.  ...  Using specific scores, dimension reduction is performed and the exact null distributions are derived for the three hypotheses.  ...  As it can be seen from the expression in (5) problems occur in high dimensions because S −1 does not exist when p > N − r(C) + 1 .  ... 
doi:10.1007/s42519-020-00154-z fatcat:ukcskrabs5anzjzckeqxppcjqa

Summarizing Complexity in High Dimensions

Karl Young, Yue Chen, John Kornak, Gerald B. Matson, Norbert Schuff
2005 Physical Review Letters  
University of California, San Francisco, USA As the need to analyze high dimensional, multispectral data on complex physical systems becomes more common, the value of methods that glean useful summary  ...  Introduction There is currently an explosion of data provided by high precision measurements in areas such as cosmology, astrophysics, high energy physics and medical imaging.  ...  A third quantity is excess entropy (EE), defined in [Feld] . EE is complementary to SC and H, and can be shown to provide important additional information.  ... 
doi:10.1103/physrevlett.94.098701 pmid:15784007 fatcat:3rihxmlww5erjizkynfcuycaf4

Robust PCA in High-dimension: A Deterministic Approach [article]

Jiashi Feng
2012 arXiv   pre-print
We consider principal component analysis for contaminated data-set in the high dimensional regime, where the dimensionality of each observation is comparable or even more than the number of observations  ...  We propose a deterministic high-dimensional robust PCA algorithm which inherits all theoretical properties of its randomized counterpart, i.e., it is tractable, robust to contaminated points, easily kernelizable  ...  This is appealing in practice, as we can terminate DHR-PCA at any time and obtain a satisfactory result in practical implementation.  ... 
arXiv:1206.4628v1 fatcat:ihilmgwbzjf4nm5rq4gnctxcxe

Averaging causal estimators in high dimensions

Joseph Antonelli, Matthew Cefalu
2020 Journal of Causal Inference  
Averaging is widely used in statistics for problems such as prediction, where there are many possible models, and averaging can improve performance and increase robustness to using incorrect models.  ...  We show that these ideas carry over into the estimation of causal effects in high-dimensional scenarios.  ...  Software An R package for implementing the averaging estimator can be found at https://github.com/jantonelli111/AveragingCausalHD.  ... 
doi:10.1515/jci-2019-0017 fatcat:g3pzgebwpvbwtlibqiqgkbg7pu

Selective Factor Extraction in High Dimensions [article]

Yiyuan She
2016 arXiv   pre-print
A class of algorithms is developed that can accommodate various convex and nonconvex sparsity-inducing penalties, and can be used for rank-constrained variable screening in high-dimensional multivariate  ...  The paper also showcases applications in macroeconomics and computer vision to demonstrate how low-dimensional data structures can be effectively captured by joint variable selection and projection.  ...  Second, we develop a computational framework with guaranteed convergence, where any thresholding rule can be applied. The algorithms adapt to reduced rank variable screening in very high dimensions.  ... 
arXiv:1403.6212v4 fatcat:x2myzxypand2zmvr77ytydalfy

Robustness May Be at Odds with Accuracy [article]

Dimitris Tsipras, Shibani Santurkar, Logan Engstrom, Alexander Turner, Aleksander Madry
2019 arXiv   pre-print
Specifically, training robust models may not only be more resource-consuming, but also lead to a reduction of standard accuracy.  ...  These differences, in particular, seem to result in unexpected benefits: the representations learned by robust models tend to align better with salient data characteristics and human perception.  ...  Dimitris Tsipras was supported in part by the NSF grant CCF-1553428. Aleksander Mądry was supported in part by an Alfred P.  ... 
arXiv:1805.12152v5 fatcat:oy4xwgaclng7th3w2worocg6za

Command-line interfaces can be efficiently brought to graphics: COLIMATE (the COmmand LIne MATE)

C. O. S. Sorzano, J. M. Carazo, O. Trelles
2002 Software, Practice & Experience  
In this work, a model for command-line-driven packages is given, and at the same time the model includes objects that can be directly translated into a graphical user interface.  ...  Furthermore, the command-line program needs not to be modified, so in this way the possibility of making batches is not lost.  ...  CONCLUSIONS In this paper we have shown how the specific structure of command-line textual interfaces can be exploited in order to produce high-quality GUIs in a very efficient manner.  ... 
doi:10.1002/spe.465 fatcat:fxmwhdkt2bb4hla7rhkg4sqfvi

Optimal shrinkage-based portfolio selection in high dimensions [article]

Taras Bodnar, Yarema Okhrin, Nestor Parolya
2021 arXiv   pre-print
In this paper we estimate the mean-variance portfolio in the high-dimensional case using the recent results from the theory of random matrices.  ...  Moreover, it is robust to deviations from normality.  ...  The assumption (A3) can be tested in practice by using Theorem 1 of Bodnar et al. (2021c) .  ... 
arXiv:1611.01958v5 fatcat:ggig2xu6kjhzrbytom6cjypxoe

A Novel Geometric Approach for Outlier Recognition in High Dimension [article]

Hu Ding, Mingquan Ye
2019 arXiv   pre-print
In this paper, we provide a novel algorithm for outlier recognition in high dimension based on the elegant geometric technique "core-set".  ...  Another advantage over the existing methods is that our algorithm can be naturally extended to handle multi-class inliers.  ...  We use PCA-VGG-0.5 feature in our experiment and the performance is very robust (see Table 6 (b)). Conclusion In this paper, we present a new approach for outlier recognition in high dimension.  ... 
arXiv:1804.09653v4 fatcat:y3qkjparjndxnev4pjbmb5mleq

Methodological requirements for valid tissue-based biomarker studies that can be used in clinical practice

Lawrence D. True
2014 Virchows Archiv  
These can be translated into assays to be used in clinical decision making.  ...  Reference is made to publicly available guidelines to improve on biomarker development in general and requirements for clinical use in particular.  ...  Acknowledgments This work was supported in part by the National Cancer Institute Pacific Northwest Prostate Cancer Specialized Program of Research Excellence (SPORE; P50 CA 097186-06)..  ... 
doi:10.1007/s00428-013-1531-0 pmid:24487786 pmcid:PMC4009398 fatcat:a6vpd3n3obexpmflvw265lcdyi
« Previous Showing results 1 — 15 out of 627,761 results