Filters








180 Hits in 10.6 sec

Small Sample Inference for Generalization Error in Classification Using the CUD Bound [article]

Eric B. Laber, Susan A. Murphy
2012 arXiv   pre-print
Confidence measures for the generalization error are crucial when small training samples are used to construct classifiers.  ...  We construct a confidence set for the generalization error by use of a smooth upper bound on the deviation between the resampled estimate and generalization error.  ...  Relevant Work Much of the work on small sample inference for the generalization error has focused on point estimates via resampling.  ... 
arXiv:1206.3274v1 fatcat:at6cjvucnrhuzg56km32imcja4

Small Sample Inference for Generalization Error in Classification Using the CUD Bound

Eric B Laber, Susan A Murphy
2008 Uncertainty in artificial intelligence : proceedings of the ... conference. Conference on Uncertainty in Artificial Intelligence  
Confidence measures for the generalization error are crucial when small training samples are used to construct classifiers.  ...  We construct a confidence set for the generalization error by use of a smooth upper bound on the deviation between the resampled estimate and generalization error.  ...  Relevant Work Much of the work on small sample inference for the generalization error has focused on point estimates via resampling.  ... 
pmid:20508837 pmcid:PMC2876736 fatcat:tlh5sl4raffetild373vjdn4au

ACCAMS: Additive Co-Clustering to Approximate Matrices Succinctly [article]

Alex Beutel, Amr Ahmed, Alexander J. Smola
2014 arXiv   pre-print
Even more surprisingly, we find that summing over small co-clusterings is more effective in modeling matrices than classic co-clustering, which uses just one large partitioning of the matrix.  ...  Following Occam's razor principle suggests that the simple structure induced by our model better captures the latent preferences and decision making processes present in the real world than classic co-clustering  ...  We would like to thank Christos Faloutsos for his valuable feedback throughout the preparation of this paper.  ... 
arXiv:1501.00199v1 fatcat:dejac7xobvccxintxr4yoywn64

Explanation-based generalization in game playing: Quantitative results [chapter]

Stefan Schrödl
1998 Lecture Notes in Computer Science  
The issues of training complexity, speedup for recognition and classification, as well as the question of optimal reasoning under time constraints are analyzed.  ...  This is due to the fact that the entire problem space is known (and can be generated) in advance. We present the main results of a large-scale empirical study.  ...  Define tQ(CUD) := hQ(1 -e(C)).t~(C) + (1 -hQ(1 -e(C))). in the case of negative examples).  ... 
doi:10.1007/bfb0026696 fatcat:u63lflkdfvb4bcpg4oj2g4csge

Admissible Numbers in the Theory of Probability

Arthur H. Copeland
1928 American Journal of Mathematics  
In combination with the error estimates in §2 (in particular, Theorem 2.3 and (2.7)), this ties in rather nicely with the probabiustic error bound for the Monte Carlo method.  ...  If we use, for instance, m = 2 32 , X = 1812433253, and r odd, then K = K(K/m) = 2, so that the bound in (11.20) becomes 65, whereas the bound in (11.19) is 50 and the bound in (11.13) is 49^.  ... 
doi:10.2307/2370609 fatcat:xbp6snpyzjeibo4d4cxlvbjonq

Quasi-Monte Carlo methods and pseudo-random numbers

Harald Niederreiter
1978 Bulletin of the American Mathematical Society  
In combination with the error estimates in §2 (in particular, Theorem 2.3 and (2.7)), this ties in rather nicely with the probabiustic error bound for the Monte Carlo method.  ...  If we use, for instance, m = 2 32 , X = 1812433253, and r odd, then K = K(K/m) = 2, so that the bound in (11.20) becomes 65, whereas the bound in (11.19) is 50 and the bound in (11.13) is 49^.  ... 
doi:10.1090/s0002-9904-1978-14532-7 fatcat:fspghfp2g5aypecvpax7gzpjoi

Anchor tuning in Faster R-CNN for measuring corn silage physical characteristics

Christoffer Bøgelund Rasmussen, Kristian Kirk, Thomas B. Moeslund
2021 Computers and Electronics in Agriculture  
We explore strategies of data sampling for specialist models, transfer learning from differing domains and tuning the anchors in the Region Proposal Network to accommodate for changes in object shapes  ...  steps or for samples to be sent to an off-site laboratory.  ...  The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.  ... 
doi:10.1016/j.compag.2021.106344 fatcat:opu7yhkarvgmjerctycdyho2d4

Exploring Halo Substructure with Giant Stars. IV. The Extended Structure of the Ursa Minor Dwarf Spheroidal Galaxy

Christopher Palma, Steven R. Majewski, Michael H. Siegel, Richard J. Patterson, James C. Ostheimer, Robert Link
2003 Astronomical Journal  
However, the photometric data on UMi stars alone does not allow us to determine if the candidate extratidal stars are now unbound or if they remain bound to the dSph within an extended dark matter halo  ...  The spatial distribution of both the UMi giant stars and the BHB stars are remarkably similar, and a large fraction of both samples of stars are found outside the tidal radius of UMi.  ...  Geisler (1984) proposed using the Washington M filter to measure the nearby stellar continuum so that the color MÀDDO51 can be used for luminosity classification.  ... 
doi:10.1086/367594 fatcat:mrygszqxazdbdhshai4ro66jum

Learning Semantically and Additively Compositional Distributional Representations

Ran Tian, Naoaki Okazaki, Kentaro Inui
2016 Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)  
and relation classification; meanwhile, DCS can guide building vectors for structured queries that can be directly executed.  ...  We show theoretical evidence that the vector compositions in our model conform to the logic of DCS.  ...  We thank the anonymous reviewers for their valuable comments.  ... 
doi:10.18653/v1/p16-1121 dblp:conf/acl/TianOI16 fatcat:swrsbjki45czrlahgl4igv7nei

Learning Semantically and Additively Compositional Distributional Representations [article]

Ran Tian and Naoaki Okazaki and Kentaro Inui
2016 arXiv   pre-print
and relation classification; meanwhile, DCS can guide building vectors for structured queries that can be directly executed.  ...  We show theoretical evidence that the vector compositions in our model conform to the logic of DCS.  ...  We thank the anonymous reviewers for their valuable comments.  ... 
arXiv:1606.02461v1 fatcat:fmwmncjkafb4xmpek5izkojsze

On the formation of spiral arms in dwarf galaxies [article]

Anatoly V. Zasov, Alexander V. Khoperskov, Natalia A. Zaitseva, Sergey S. Khrapov
2021 arXiv   pre-print
In this work we analyze the sample of about 40 dS-galaxies found by scanning by eye the images of late-type galaxies with m_B<15^m and M_B>-18^m and photometric diameter D_25<12 kpc.  ...  Therefore, only a small fraction of dwarf galaxies with stellar/gaseous discs have spiral or ring structures. The thicker stellar disc, the more gas is required for the spiral structure to form.  ...  ACKNOWLEDGEMENTS The authors thank the antonymous referee for fruitful remarks. NZ was supported by Russian Foundation for Basic Research (project 20-02-00080 A).  ... 
arXiv:2108.04624v1 fatcat:gvomtas5ajas7ifbh3w5eluybq

The high-resolution, three-dimensional solution structure of human interleukin-4 determined by multidimensional heteronuclear magnetic resonance spectroscopy

Robert Powers, Daniel S. Garrett, Carl J. March, Eric A. Frieden, Angela M. Gronenborn, G. Marius Clore
1993 Biochemistry  
for the backbone atoms, 0.83 f 0.03 A for all atoms, and 0.51 f 0.04 A for all atoms excluding disordered side chains.  ...  The linker elements between the helices are formed by either long loops, small helical turns, or short strands. The latter include a mini anti-parallel P-sheet.  ...  be inferred by manipulation of the functional groups using interactive modeling tools.  ... 
doi:10.1021/bi00077a030 pmid:8329398 fatcat:c7vrroyt4zaflpng2pyph4vreu

Predicting the need for vehicle compressor repairs using maintenance records and logged vehicle data

Rune Prytz, Sławomir Nowaczyk, Thorsteinn Rögnvaldsson, Stefan Byttner
2015 Engineering applications of artificial intelligence  
The machine learning based features outperform the human expert features, wich supports the idea to use data mining to improve maintenance operations in this domain.  ...  Methods and results are presented for applying supervised machine learning techniques to the task of predicting the need for repairs of air compressors in commercial trucks and buses.  ...  Acknowledgment The authors thank Vinnova (Swedish Governmental Agency for Innovation Systems), AB Volvo, Halmstad University, and the Swedish Knowledge Foundation for financial support for doing this research  ... 
doi:10.1016/j.engappai.2015.02.009 fatcat:oeezxsy3zrhwpif33p7sc6wtui

B, D and K decays [chapter]

G. Buchalla, T. K. Komatsubara, F. Muheim, L. Silvestrini, M. Artuso, D. M. Asner, P. Ball, E. Baracchini, G. Bell, M. Beneke, J. Berryhill, A. Bevan (+114 others)
2009 Flavor in the Era of the LHC  
The experimental prospects for flavour physics at future facilities are discussed in Sec. 4.  ...  Sec. 2 contains a concise discussion of the main theoretical problem in flavour physics: the evaluation of the relevant hadronic matrix elements for weak decays.  ...  The authors of this report thank all further participants in this workshop series for their presentations during the five workshop meetings, for contributing to the discussions, and for useful comments  ... 
doi:10.1007/978-3-540-95942-7_3 fatcat:2jswhezgo5bitlnojqieuc3lyq

B, D and K decays

G. Buchalla, T. K. Komatsubara, F. Muheim, L. Silvestrini, M. Artuso, D. M. Asner, P. Ball, E. Baracchini, G. Bell, M. Beneke, J. Berryhill, A. Bevan (+114 others)
2008 European Physical Journal C: Particles and Fields  
In particular, precision measurements and computations in the realm of flavour physics are expected to play a key role in constraining the unknown parameters of the Lagrangian of any NP model emerging  ...  The aim of Working Group 2 was twofold: on one hand, to provide a coherent, up-to-date picture of the status of flavour physics before the start of the LHC; on the other hand, to initiate activities on  ...  The authors of this report thank all further participants in this workshop series for their presentations during the five workshop meetings, for contributing to the discussions, and for useful comments  ... 
doi:10.1140/epjc/s10052-008-0716-1 fatcat:iyjthhovdfaoneyxc7tr6n6a74
« Previous Showing results 1 — 15 out of 180 results