A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2021; you can also visit the original URL.
The file type is application/pdf.
Filters
On the Complexity of Labeled Datasets
[article]
2021
arXiv
pre-print
Preserved Fulltext
Other Versions
The Statistical Learning Theory (SLT) provides the foundation to ensure that a supervised algorithm generalizes the mapping f: 𝒳→𝒴 given f is selected from its search space bias ℱ. ...
However, the analytical solution of the Shattering coefficient is still an open problem since the first studies by Vapnik and Chervonenkis in 1962, which we address on specific datasets, in this paper, ...
Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of Itaú Unibanco SA, CNPq, nor FAPESP. ...
arXiv:1911.05461v3
fatcat:lb2gcscd2vcrnmlih36ogcfwjy
Computing the Shattering Coefficient of Supervised Learning Algorithms
[article]
2018
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2020; you can also visit the original URL.
The file type is application/pdf.
Other Versions
This paper proves the Shattering coefficient for any Hilbert space H containing the input space X and discusses its effects in terms of learning guarantees for supervised machine algorithms. ...
The Statistical Learning Theory (SLT) provides the theoretical guarantees for supervised machine learning based on the Empirical Risk Minimization Principle (ERMP). ...
SHATTERING COEFFICIENT The Shattering coefficient N (F, 2n) of any h-dimensional Hilbert space H being classified with a single (h − 1)- ...
arXiv:1805.02627v4
fatcat:bbglwnkidbb7lkkgzxer3a3rqu
Data driven semi-supervised learning
[article]
2021
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2021; you can also visit the original URL.
The file type is application/pdf.
Other Versions
We expect some of the tools and techniques we develop along the way to be of interest beyond semi-supervised learning, for data driven algorithms for combinatorial problems more generally. ...
Over the past decades, several elegant graph-based semi-supervised learning algorithms for how to infer the labels of the unlabeled examples given the graph and a few labeled examples have been proposed ...
We provide asymptotically tight upper and lower bounds on the pseudodimension of learning the best parameter from a parameterized family of semi-supervised learning algorithms, each algorithm corresponding ...
arXiv:2103.10547v4
fatcat:tqvet3eg2nbnnmrlq7tinsofse
The Learnability of Unknown Quantum Measurements
[article]
2015
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2020; you can also visit the original URL.
The file type is application/pdf.
Our main result in the paper is that, for learning an unknown quantum measurement, the upper bound, given by the fat-shattering dimension, is linearly proportional to the dimension of the underlying Hilbert ...
This kind of problem is, arguably, one of the most fundamental problems in statistical learning theory and the bounds for practical settings can be completely characterised by a simple measure of complexity ...
procedure of supervised, unsupervised and SVM algorithms. ...
arXiv:1501.00559v1
fatcat:5wtqou5pi5e6hgcu54n4rphyce
A Survey on Semi-Supervised Learning for Delayed Partially Labelled Data Streams
[article]
2021
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2021; you can also visit the original URL.
The file type is application/pdf.
To address the learning problems associated with such data, one can ignore the unlabelled data and focus only on the labelled data (supervised learning); use the labelled data and attempt to leverage the ...
unlabelled data (semi-supervised learning); or assume some labels will be available on request (active learning). ...
In that sense, let us assume the Shattering coefficient function N (F , 2 ) = 2 for a specific semi-supervised algorithm working on some -dimensional Hilbert input space, thus defining the maximal number ...
arXiv:2106.09170v1
fatcat:nn4ja4ptsndwngyocjne2tp4ae
Activized Learning: Transforming Passive to Active with Improved Label Complexity
[article]
2011
arXiv
pre-print
Preserved Fulltext
The Internet Archive has a preservation copy of this work in our general collections.
The file type is application/pdf.
We further provide a general characterization of the magnitudes of these improvements in terms of a novel generalization of the disagreement coefficient. ...
We study the theoretical advantages of active learning over passive learning. ...
Background The term active learning refers to a family of supervised learning protocols, characterized by the ability of the learning algorithm to pose queries to a teacher, who has access to the target ...
arXiv:1108.1766v1
fatcat:hldmtf567vdg5aa7zjpllk4n4u
A general agnostic active learning algorithm
2007
Neural Information Processing Systems
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2021; you can also visit the original URL.
The file type is application/pdf.
Our algorithm extends the simple scheme of Cohn, Atlas, and Ladner [1] to the agnostic setting, using reductions to supervised learning that harness generalization bounds in a simple but subtle manner. ...
We present an agnostic active learning algorithm for any hypothesis class of bounded VC dimension under arbitrary data distributions. ...
ensures the nth shatter coefficient S(H, n) is at most O(n d ) by Sauer's lemma). ...
dblp:conf/nips/DasguptaHM07
fatcat:3dy4fwfjq5fopnbtt6f6ckvfou
Activized Learning with Uniform Classification Noise
2013
International Conference on Machine Learning
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2022; you can also visit the original URL.
The file type is application/pdf.
We prove that for any VC class, it is possible to transform any passive learning algorithm into an active learning algorithm with strong asymptotic improvements in label complexity for every nontrivial ...
This generalizes a similar result proven by (Hanneke, 2009; 2012) for the realizable case, and is the first result establishing that such general improvement guarantees are possible in the presence of ...
It is therefore important to try to reduce the total number of labels needed for supervised learning. ...
dblp:conf/icml/YangH13
fatcat:w4ye3ymbrrdtxlky6vgsfv66ka
Theoretical learning guarantees applied to acoustic modeling
2019
Journal of the Brazilian Computer Society
Preserved Fulltext
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 application/pdf.
Furthermore, we present proof that our algorithm does adhere to the learning guarantees provided by the statistical learning theory. ...
More importantly, we isolate the performance of the acoustic model and provide results on both the frame and phoneme level, considering the true robustness of the model. ...
Also, we thank Capes for the grant PROEX-7901561/D to MF. ...
doi:10.1186/s13173-018-0081-3
fatcat:vdxoozswkjfznkhwyfjpt3runy
The VC-Dimension of Similarity Hypotheses Spaces
[article]
2015
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2020; you can also visit the original URL.
The file type is application/pdf.
Given a set X and a function h:X⟶{0,1} which labels each element of X with either 0 or 1, we may define a function h^(s) to measure the similarity of pairs of points in X according to h. ...
This idea can be extended to a set of functions, or hypothesis space H⊆{0,1}^X by defining a similarity hypothesis space H^(s):={h^(s):h∈H}. We show that vc-dimension(H^(s)) ∈Θ(vc-dimension(H)). ...
In particular we would like thank Ruth Urner for proving an initial motivating upper bound of d(H (s) ) ≤ 2d(H) log(2d(H)). ...
arXiv:1502.07143v1
fatcat:utlfcoh6dja5hevnciaue64zla
Finite Sample Complexity of Rare Pattern Anomaly Detection
2016
Conference on Uncertainty in Artificial Intelligence
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2022; you can also visit the original URL.
The file type is application/pdf.
However, compared to supervised learning, there has been very little work aimed at understanding the sample complexity of anomaly detection. ...
In analogy with the PAC framework for supervised learning, we develop sample complexity results that relate the complexity of the pattern space to the data requirements needed for PAC guarantees. ...
Acknowledgements This work is partially supported by the Future of Life Institute and DARPA under contract number FA8650-15-C-7557 and W911NF-11-C-0088. ...
dblp:conf/uai/SiddiquiFDD16
fatcat:d2ikhytdgvbijfqtvt6qeeeyuu
Bounds for the VC Dimension of 1NN Prototype Sets
[article]
2019
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2020; you can also visit the original URL.
The file type is application/pdf.
We discuss the implications of these bounds for the size of training set needed to learn such a classifier to a given accuracy. ...
In Statistical Learning, the Vapnik-Chervonenkis (VC) dimension is an important combinatorial property of classifiers. ...
While preparing the paper for publication, IG received support from the European Union's Horizon 2020 research and Innovation programme under grant agreement No 731593. ...
arXiv:1902.02660v1
fatcat:ikfbxwkrtbao7itqd6csfsrsuq
Approximation beats concentration? An approximation view on inference with smooth radial kernels
[article]
2018
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2020; you can also visit the original URL.
The file type is application/pdf.
Other Versions
We also investigate the fitting capacity of kernels, giving explicit bounds on the fat shattering dimension of the balls in Reproducing Kernel Hilbert spaces. ...
We analyze eigenvalue decay of kernels operators and matrices, properties of eigenfunctions/eigenvectors and "Fourier" coefficients of functions in the kernel space restricted to a discrete set of data ...
He would like to acknowledge NSF funding and the hospitality of the Simons Institute for the Theory of Computing, where early discussions took place. ...
arXiv:1801.03437v2
fatcat:bp3imztybzhg7c6qu3nj7htyjq
Robust Phoneme Recognition with Little Data
2019
Symposium on Languages, Applications and Technologies
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2022; you can also visit the original URL.
The file type is application/pdf.
and calculating the shattering coefficient for the algorithms used. ...
We also beat the best replication study of the state of the art with a 28% FER. ...
Additionally, Vapnik proved a bound for supervised learning algorithms considering the shattering coefficient N (F, 2n). ...
doi:10.4230/oasics.slate.2019.4
dblp:conf/slate/ShulbyFMA19
fatcat:lesl5ro5pjh3beo5i4ewvhg3qm
PAC-learning in the presence of evasion adversaries
[article]
2018
arXiv
pre-print
Preserved Fulltext
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 application/pdf.
Other Versions
The existence of evasion attacks during the test phase of machine learning algorithms represents a significant challenge to both their deployment and understanding. ...
In this paper, we step away from the attack-defense arms race and seek to understand the limits of what can be learned in the presence of an evasion adversary. ...
In this paper, we take a more fundamental approach to understanding the robustness of supervised classification algorithms by extending well-understood results for supervised batch learning in statistical ...
arXiv:1806.01471v2
fatcat:nvwuhgloqzhp5hckltscsfo4gi
« Previous
Showing results 1 — 15 out of 678 results