Results 1  10
of
46
Approximate Statistical Tests for Comparing Supervised Classification Learning Algorithms
, 1998
"... This article reviews five approximate statistical tests for determining whether one learning algorithm outperforms another on a particular learning task. These tests are compared experimentally to determine their probability of incorrectly detecting a difference when no difference exists (type I err ..."
Abstract

Cited by 531 (8 self)
 Add to MetaCart
This article reviews five approximate statistical tests for determining whether one learning algorithm outperforms another on a particular learning task. These tests are compared experimentally to determine their probability of incorrectly detecting a difference when no difference exists (type I error). Two widely used statistical tests are shown to have high probability of type I error in certain situations and should never be used: a test for the difference of two proportions and a paireddifferences t test based on taking several random traintest splits. A third test, a paireddifferences t test based on 10fold crossvalidation, exhibits somewhat elevated probability of type I error. A fourth test, McNemar’s test, is shown to have low type I error. The fifth test is a new test, 5 × 2 cv, based on five iterations of twofold crossvalidation. Experiments show that this test also has acceptable type I error. The article also measures the power (ability to detect algorithm differences when they do exist) of these tests. The crossvalidated t test is the most powerful. The 5×2 cv test is shown to be slightly more powerful than McNemar’s test. The choice of the best test is determined by the computational cost of running the learning algorithm. For algorithms that can be executed only once, McNemar’s test is the only test with acceptable type I error. For algorithms that can be executed 10 times, the 5×2 cv test is recommended, because it is slightly more powerful and because it directly measures variation due to the choice of training set.
A Guide to the Literature on Learning Probabilistic Networks From Data
, 1996
"... This literature review discusses different methods under the general rubric of learning Bayesian networks from data, and includes some overlapping work on more general probabilistic networks. Connections are drawn between the statistical, neural network, and uncertainty communities, and between the ..."
Abstract

Cited by 172 (0 self)
 Add to MetaCart
This literature review discusses different methods under the general rubric of learning Bayesian networks from data, and includes some overlapping work on more general probabilistic networks. Connections are drawn between the statistical, neural network, and uncertainty communities, and between the different methodological communities, such as Bayesian, description length, and classical statistics. Basic concepts for learning and Bayesian networks are introduced and methods are then reviewed. Methods are discussed for learning parameters of a probabilistic network, for learning the structure, and for learning hidden variables. The presentation avoids formal definitions and theorems, as these are plentiful in the literature, and instead illustrates key concepts with simplified examples. Keywords Bayesian networks, graphical models, hidden variables, learning, learning structure, probabilistic networks, knowledge discovery. I. Introduction Probabilistic networks or probabilistic gra...
An experimental and theoretical comparison of model selection methods. Machine Learning 27
, 1997
"... In the model selection problem, we must balance the complexity of a statistical model with its goodness of fit to the training data. This problem arises repeatedly in statistical estimation, machine learning, and scientific inquiry in general. ..."
Abstract

Cited by 110 (5 self)
 Add to MetaCart
In the model selection problem, we must balance the complexity of a statistical model with its goodness of fit to the training data. This problem arises repeatedly in statistical estimation, machine learning, and scientific inquiry in general.
Wrappers For Performance Enhancement And Oblivious Decision Graphs
, 1995
"... In this doctoral dissertation, we study three basic problems in machine learning and two new hypothesis spaces with corresponding learning algorithms. The problems we investigate are: accuracy estimation, feature subset selection, and parameter tuning. The latter two problems are related and are stu ..."
Abstract

Cited by 107 (8 self)
 Add to MetaCart
In this doctoral dissertation, we study three basic problems in machine learning and two new hypothesis spaces with corresponding learning algorithms. The problems we investigate are: accuracy estimation, feature subset selection, and parameter tuning. The latter two problems are related and are studied under the wrapper approach. The hypothesis spaces we investigate are: decision tables with a default majority rule (DTMs) and oblivious readonce decision graphs (OODGs).
Algorithmic Stability and SanityCheck Bounds for LeaveOneOut CrossValidation
 Neural Computation
, 1997
"... In this paper we prove sanitycheck bounds for the error of the leaveoneout crossvalidation estimate of the generalization error: that is, bounds showing that the worstcase error of this estimate is not much worse than that of the training error estimate. The name sanitycheck refers to the fact ..."
Abstract

Cited by 100 (0 self)
 Add to MetaCart
In this paper we prove sanitycheck bounds for the error of the leaveoneout crossvalidation estimate of the generalization error: that is, bounds showing that the worstcase error of this estimate is not much worse than that of the training error estimate. The name sanitycheck refers to the fact that although we often expect the leaveoneout estimate to perform considerably better than the training error estimate, we are here only seeking assurance that its performance will not be considerably worse. Perhaps surprisingly, such assurance has been given only for limited cases in the prior literature on crossvalidation. Any nontrivial bound on the error of leaveoneout must rely on some notion of algorithmic stability. Previous bounds relied on the rather strong notion of hypothesis stability, whose application was primarily limited to nearestneighbor and other local algorithms. Here we introduce the new and weaker notion of error stability, and apply it to obtain sanitycheck b...
Efficient Progressive Sampling
, 1999
"... Having access to massiveamounts of data does not necessarily imply that induction algorithms must use them all. Samples often provide the same accuracy with far less computational cost. However, the correct sample size is rarely obvious. We analyze methods for progressive samplingstarting with ..."
Abstract

Cited by 91 (9 self)
 Add to MetaCart
Having access to massiveamounts of data does not necessarily imply that induction algorithms must use them all. Samples often provide the same accuracy with far less computational cost. However, the correct sample size is rarely obvious. We analyze methods for progressive samplingstarting with small samples and progressively increasing them as long as model accuracy improves. We show that a simple, geometric sampling schedule is efficient in an asymptotic sense. We then explore the notion of optimal efficiency: what is the absolute best sampling schedule? We describe the issues involved in instantiating an "optimally efficient" progressive sampler. Finally,we provide empirical results comparing a variety of progressive sampling methods. We conclude that progressive sampling often is preferable to analyzing all data instances.
PACBayesian Model Averaging
 In Proceedings of the Twelfth Annual Conference on Computational Learning Theory
, 1999
"... PACBayesian learning methods combine the informative priors of Bayesian methods with distributionfree PAC guarantees. Building on earlier methods for PACBayesian model selection, this paper presents a method for PACBayesian model averaging. The main result is a bound on generalization error of a ..."
Abstract

Cited by 75 (2 self)
 Add to MetaCart
PACBayesian learning methods combine the informative priors of Bayesian methods with distributionfree PAC guarantees. Building on earlier methods for PACBayesian model selection, this paper presents a method for PACBayesian model averaging. The main result is a bound on generalization error of an arbitrary weighted mixture of concepts that depends on the empirical error of that mixture and the KLdivergence of the mixture from the prior. A simple characterization is also given for the error bound achieved by the optimal weighting. 1
Tree induction vs. logistic regression: A learningcurve analysis
 CEDER WORKING PAPER #IS0102, STERN SCHOOL OF BUSINESS
, 2001
"... Tree induction and logistic regression are two standard, offtheshelf methods for building models for classi cation. We present a largescale experimental comparison of logistic regression and tree induction, assessing classification accuracy and the quality of rankings based on classmembership pr ..."
Abstract

Cited by 62 (16 self)
 Add to MetaCart
Tree induction and logistic regression are two standard, offtheshelf methods for building models for classi cation. We present a largescale experimental comparison of logistic regression and tree induction, assessing classification accuracy and the quality of rankings based on classmembership probabilities. We use a learningcurve analysis to examine the relationship of these measures to the size of the training set. The results of the study show several remarkable things. (1) Contrary to prior observations, logistic regression does not generally outperform tree induction. (2) More specifically, and not surprisingly, logistic regression is better for smaller training sets and tree induction for larger data sets. Importantly, this often holds for training sets drawn from the same domain (i.e., the learning curves cross), so conclusions about inductionalgorithm superiority on a given domain must be based on an analysis of the learning curves. (3) Contrary to conventional wisdom, tree induction is effective atproducing probabilitybased rankings, although apparently comparatively less so foragiven training{set size than at making classifications. Finally, (4) the domains on which tree induction and logistic regression are ultimately preferable canbecharacterized surprisingly well by a simple measure of signaltonoise ratio.
The evolutionary dynamics of grammar acquisition
 J. THEOR. BIOL
, 2001
"... Grammar is the computational system of language. It is a set of rules that speci"es how to construct sentences out of words. Grammar is the basis of the unlimited expressibility of human language. Children acquire the grammar of their native language without formal education simply by hearing a numb ..."
Abstract

Cited by 32 (6 self)
 Add to MetaCart
Grammar is the computational system of language. It is a set of rules that speci"es how to construct sentences out of words. Grammar is the basis of the unlimited expressibility of human language. Children acquire the grammar of their native language without formal education simply by hearing a number of sample sentences. Children could not solve this learning task if they did not have some preformed expectations. In other words, children have to evaluate the sample sentences and choose one grammar out of a limited set of candidate grammars. The restricted search space and the mechanism which allows to evaluate the sample sentences is called universal grammar. Universal grammar cannot be learned; it must be in place when the learning process starts. In this paper, we design a mathematical theory that places the problem of language acquisition into an evolutionary context. We formulate equations for the population dynamics of communication and grammar learning. We ask how accurate children have to learn the grammar of their parents ' language for a population of individuals to evolve and maintain a coherent grammatical system. It turns out that there is a maximum error tolerance for which a predominant grammar is stable. We calculate the maximum size of the search space that is compatible with coherent communication in a population. Thus, we specify the conditions for the evolution of universal grammar.
Predictability, Complexity, and Learning
, 2001
"... We define predictive information Ipred(T) as the mutual information between the past and the future of a time series. Three qualitatively different behaviors are found in the limit of large observation times T: Ipred(T) can remain finite, grow logarithmically, or grow as a fractional power law. If t ..."
Abstract

Cited by 30 (2 self)
 Add to MetaCart
We define predictive information Ipred(T) as the mutual information between the past and the future of a time series. Three qualitatively different behaviors are found in the limit of large observation times T: Ipred(T) can remain finite, grow logarithmically, or grow as a fractional power law. If the time series allows us to learn a model with a finite number of parameters, then Ipred(T) grows logarithmically with a coefficient that counts the dimensionality of the model space. In contrast, powerlaw growth is associated, for example, with the learning of infinite parameter (or nonparametric) models such as continuous functions with smoothness constraints. There are connections between the predictive information and measures of complexity that have been defined both in learning theory and the analysis of physical systems through statistical mechanics and dynamical systems theory. Furthermore, in the same way that entropy provides the unique measure of available information consistent with some simple and plausible conditions, we argue that the divergent part of Ipred(T) provides the unique measure for the complexity of dynamics underlying a time series. Finally, we discuss how these ideas may be useful in problems in physics, statistics, and biology.