ICA as a preprocessing technique for classification

In this paper we propose the use of the independent component analysis (ICA) [1] technique for improving the classification rate of decision trees and multilayer perceptrons [2], [3]. The use of an ICA for the preprocessing stage, makes the structure of both classifiers simpler, and therefore improves the generalization properties. The hypothesis behind the proposed preprocessing is that an ICA analysis will transform the feature space into a space where the components are independent, and aligned to the axes and therefore will be more adapted to the way that a decision tree is constructed. Also the inference of the weights of a multilayer perceptron will be much easier because the gradient search in the weight space will follow independent trajectories. The result is that classifiers are less complex and on some databases the error rate is lower. This idea is also applicable to regression

Power supply noise and logic error probability

Voltage fluctuations caused by parasitic impedances in the power supply rails of modern ICs are a major concern in nowadays ICs. The voltage fluctuations are spread out to the diverse nodes of the internal sections causing two effects: a degradation of performances mainly impacting gate delays anda noisy contamination of the quiescent levels of the logic that drives the node. Both effects are presented together, in thispaper, showing than both are a cause of errors in modern and future digital circuits. The paper groups both error mechanismsand shows how the global error rate is related with the voltage deviation and the period of the clock of the digital system.

A systolic array for linear MIMO detection based on an all-swap lattice reduction algorithm

A systolic array to implement lattice-reduction-aided lineardetection is proposed for a MIMO receiver. The lattice reductionalgorithm and the ensuing linear detections are operated in the same array, which can be hardware-efficient. All-swap lattice reduction algorithm (ASLR) is considered for the systolic design.ASLR is a variant of the LLL algorithm, which processes all lattice basis vectors within one iteration. Lattice-reduction-aided linear detection based on ASLR and LLL algorithms have very similarbit-error-rate performance, while ASLR is more time efficient inthe systolic array, especially for systems with a large number ofantennas.

Stepwise multiple testing as formalized data snooping

It is common in econometric applications that several hypothesis tests arecarried out at the same time. The problem then becomes how to decide whichhypotheses to reject, accounting for the multitude of tests. In this paper,we suggest a stepwise multiple testing procedure which asymptoticallycontrols the familywise error rate at a desired level. Compared to relatedsingle-step methods, our procedure is more powerful in the sense that itoften will reject more false hypotheses. In addition, we advocate the useof studentization when it is feasible. Unlike some stepwise methods, ourmethod implicitly captures the joint dependence structure of the teststatistics, which results in increased ability to detect alternativehypotheses. We prove our method asymptotically controls the familywise errorrate under minimal assumptions. We present our methodology in the context ofcomparing several strategies to a common benchmark and deciding whichstrategies actually beat the benchmark. However, our ideas can easily beextended and/or modied to other contexts, such as making inference for theindividual regression coecients in a multiple regression framework. Somesimulation studies show the improvements of our methods over previous proposals. We also provide an application to a set of real data.

Measuring strategic uncertainty in coordination games

This paper explores three aspects of strategic uncertainty: its relation to risk, predictability of behavior and subjective beliefs of players. In a laboratory experiment we measure subjects certainty equivalents for three coordination games and one lottery. Behavior in coordination games is related to risk aversion, experience seeking, and age.From the distribution of certainty equivalents we estimate probabilities for successful coordination in a wide range of games. For many games, success of coordination is predictable with a reasonable error rate. The best response to observed behavior is close to the global-game solution. Comparing choices in coordination games with revealed risk aversion, we estimate subjective probabilities for successful coordination. In games with a low coordination requirement, most subjects underestimate the probability of success. In games with a high coordination requirement, most subjects overestimate this probability. Estimating probabilistic decision models, we show that the quality of predictions can be improved when individual characteristics are taken into account. Subjects behavior is consistent with probabilistic beliefs about the aggregate outcome, but inconsistent with probabilistic beliefs about individual behavior.

Exact and approximate stepdown methods for multiple hypothesis testing

Consider the problem of testing k hypotheses simultaneously. In this paper,we discuss finite and large sample theory of stepdown methods that providecontrol of the familywise error rate (FWE). In order to improve upon theBonferroni method or Holm's (1979) stepdown method, Westfall and Young(1993) make eective use of resampling to construct stepdown methods thatimplicitly estimate the dependence structure of the test statistics. However,their methods depend on an assumption called subset pivotality. The goalof this paper is to construct general stepdown methods that do not requiresuch an assumption. In order to accomplish this, we take a close look atwhat makes stepdown procedures work, and a key component is a monotonicityrequirement of critical values. By imposing such monotonicity on estimatedcritical values (which is not an assumption on the model but an assumptionon the method), it is demonstrated that the problem of constructing a validmultiple test procedure which controls the FWE can be reduced to the problemof contructing a single test which controls the usual probability of a Type 1error. This reduction allows us to draw upon an enormous resamplingliterature as a general means of test contruction.

Validation procedures in radiological diagnostic models. Neural network and logistic regression

The objective of this paper is to compare the performance of twopredictive radiological models, logistic regression (LR) and neural network (NN), with five different resampling methods. One hundred and sixty-seven patients with proven calvarial lesions as the only known disease were enrolled. Clinical and CT data were used for LR and NN models. Both models were developed with cross validation, leave-one-out and three different bootstrap algorithms. The final results of each model were compared with error rate and the area under receiver operating characteristic curves (Az). The neural network obtained statistically higher Az than LR with cross validation. The remaining resampling validation methods did not reveal statistically significant differences between LR and NN rules. The neural network classifier performs better than the one based on logistic regression. This advantage is well detected by three-fold cross-validation, but remains unnoticed when leave-one-out or bootstrap algorithms are used.

Statistics of transverse mode turn-on dynamics in VCSEL's

The turn-on process of a multimode VCSEL is investigated from a statistical point of view. Special attention is paid to quantities such as time jitter and bit error rate. The single-mode performance of VCSEL¿s during current modulation is compared to that of edge-emitting lasers.

To select or to weigh: a comparative study of linear combination schemes for superparent-one-dependence estimators

We conduct a large-scale comparative study on linearly combining superparent-one-dependence estimators (SPODEs), a popular family of seminaive Bayesian classifiers. Altogether, 16 model selection and weighing schemes, 58 benchmark data sets, and various statistical tests are employed. This paper's main contributions are threefold. First, it formally presents each scheme's definition, rationale, and time complexity and hence can serve as a comprehensive reference for researchers interested in ensemble learning. Second, it offers bias-variance analysis for each scheme's classification error performance. Third, it identifies effective schemes that meet various needs in practice. This leads to accurate and fast classification algorithms which have an immediate and significant impact on real-world applications. Another important feature of our study is using a variety of statistical tests to evaluate multiple learning methods across multiple data sets.

Shallow-transfer rule-based machine translation for Czech to Polish

This article describes the developmentof an Open Source shallow-transfer machine translation system from Czech to Polish in theApertium platform. It gives details ofthe methods and resources used in contructingthe system. Although the resulting system has quite a high error rate, it is still competitive with other systems.

High-SNR analytical performance of spatial multiplexing MIMO systems with CSI

In this paper, we investigate the average andoutage performance of spatial multiplexing multiple-input multiple-output (MIMO) systems with channel state information at both sides of the link. Such systems result, for example, from exploiting the channel eigenmodes in multiantenna systems. Dueto the complexity of obtaining the exact expression for the average bit error rate (BER) and the outage probability, we deriveapproximations in the high signal-to-noise ratio (SNR) regime assuming an uncorrelated Rayleigh flat-fading channel. Moreexactly, capitalizing on previous work by Wang and Giannakis, the average BER and outage probability versus SNR curves ofspatial multiplexing MIMO systems are characterized in terms of two key parameters: the array gain and the diversity gain. Finally, these results are applied to analyze the performance of a variety of linear MIMO transceiver designs available in the literature.

Robust power allocation algorithms for MIMO OFDM systems with imperfect CSI

This paper presents a Bayesian approach to the design of transmit prefiltering matrices in closed-loop schemes robust to channel estimation errors. The algorithms are derived for a multiple-input multiple-output (MIMO) orthogonal frequency division multiplexing (OFDM) system. Two different optimizationcriteria are analyzed: the minimization of the mean square error and the minimization of the bit error rate. In both cases, the transmitter design is based on the singular value decomposition (SVD) of the conditional mean of the channel response, given the channel estimate. The performance of the proposed algorithms is analyzed,and their relationship with existing algorithms is indicated. As withother previously proposed solutions, the minimum bit error rate algorithmconverges to the open-loop transmission scheme for very poor CSI estimates.

Diagnosis of Alzheimer’s Disease from EEG by Means of Synchrony Measures in Optimized Frequency Bands

Several clinical studies have reported that EEG synchrony is affected by Alzheimer’s disease (AD). In this paper a frequency band analysis of AD EEG signals is presented, with the aim of improving the diagnosis of AD using EEG signals. In this paper, multiple synchrony measures are assessed through statistical tests (Mann–Whitney U test), including correlation, phase synchrony and Granger causality measures. Moreover, linear discriminant analysis (LDA) is conducted with those synchrony measures as features. For the data set at hand, the frequency range (5-6Hz) yields the best accuracy for diagnosing AD, which lies within the classical theta band (4-8Hz). The corresponding classification error is 4.88% for directed transfer function (DTF) Granger causality measure. Interestingly, results show that EEG of AD patients is more synchronous than in healthy subjects within the optimized range 5-6Hz, which is in sharp contrast with the loss of synchrony in AD EEG reported in many earlier studies. This new finding may provide new insights about the neurophysiology of AD. Additional testing on larger AD datasets is required to verify the effectiveness of the proposed approach.

Adaptive model selection using empirical complexities

Given $n$ independent replicates of a jointly distributed pair $(X,Y)\in {\cal R}^d \times {\cal R}$, we wish to select from a fixed sequence of model classes ${\cal F}_1, {\cal F}_2, \ldots$ a deterministic prediction rule $f: {\cal R}^d \to {\cal R}$ whose risk is small. We investigate the possibility of empirically assessingthe {\em complexity} of each model class, that is, the actual difficulty of the estimation problem within each class. The estimated complexities are in turn used to define an adaptive model selection procedure, which is based on complexity penalized empirical risk.The available data are divided into two parts. The first is used to form an empirical cover of each model class, and the second is used to select a candidate rule from each cover based on empirical risk. The covering radii are determined empirically to optimize a tight upper bound on the estimation error. An estimate is chosen from the list of candidates in order to minimize the sum of class complexity and empirical risk. A distinguishing feature of the approach is that the complexity of each model class is assessed empirically, based on the size of its empirical cover.Finite sample performance bounds are established for the estimates, and these bounds are applied to several non-parametric estimation problems. The estimates are shown to achieve a favorable tradeoff between approximation and estimation error, and to perform as well as if the distribution-dependent complexities of the model classes were known beforehand. In addition, it is shown that the estimate can be consistent,and even possess near optimal rates of convergence, when each model class has an infinite VC or pseudo dimension.For regression estimation with squared loss we modify our estimate to achieve a faster rate of convergence.

Model selection and error estimation

We study model selection strategies based on penalized empirical loss minimization. We point out a tight relationship between error estimation and data-based complexity penalization: any good error estimate may be converted into a data-based penalty function and the performance of the estimate is governed by the quality of the error estimate. We consider several penalty functions, involving error estimates on independent test data, empirical {\sc vc} dimension, empirical {\sc vc} entropy, andmargin-based quantities. We also consider the maximal difference between the error on the first half of the training data and the second half, and the expected maximal discrepancy, a closely related capacity estimate that can be calculated by Monte Carlo integration. Maximal discrepancy penalty functions are appealing for pattern classification problems, since their computation is equivalent to empirical risk minimization over the training data with some labels flipped.