Sequential non-stationary dynamic classification with sparse feedback

Many data analysis problems require robust tools for discerning between states or classes in the data. In this paper we consider situations in which the decision boundaries between classes are potentially non-linear and subject to “concept drift” and hence static classifiers fail. The applications for which we present results are characterized by the requirement that robust online decisions be made and by the fact that target labels may be missing, so there is very often no feedback regarding the system's performance. The inherent non-stationarity in the data is tracked using a non-linear dynamic classifier, the parameters of which evolve under an extended Kalman filter framework, derived using a sequential Bayesian-learning paradigm. The method is extended to take into account missing and incorrectly labeled targets and to actively request target labels. The method is shown to work well in simulation as well as when applied to sequential decision problems in medical signal analysis.     

FRVT 2006: Quo Vadis face quality

A study is presented showing how three state-of-the-art algorithms from the Face Recognition Vendor Test 2006 (FRVT 2006) are effected by factors related to face images and people. The recognition scenario compares highly controlled images to images taken of people as they stand before a camera in settings such as hallways and outdoors in front of buildings. A Generalized Linear Mixed Model (GLMM) is used to estimate the probability an algorithm successfully verifies a person conditioned upon the factors included in the study. The factors associated with people are: Gender, Race, Age and whether they wear Glasses. The factors associated with images are: the size of the face, edge density and region density. The setting, indoors versus outdoors, is also a factor. Edge density can change the estimated probability of verification dramatically, for example from about 0.15 to 0.85. However, this effect is not consistent across algorithm or setting. This finding shows that simple measurable factors are capable of characterizing face quality; however, these factors typically interact with both algorithm and setting.     

Predictive-collaborative model as recovery and validation tool. Case of study: Psychiatric emergency department decision support

There is a frequent situation in data mining where data collected must be used in real time to support decisions and they could present missing or non consistent values. The objective of this proposal consists of the recovery of missing values and verifies the consistency and integrity of the provided, in order to increase the information to support decisions. To address this, a predictive-collaborative model has been designed. It is composed of different predictive models generated by means of a training set and classifier selection algorithm. The combined suggestions of these predictive models are offered to support decisions. As case of study, the psychiatric emergency department at the Doce de Octubre Hospital in Madrid has been considered, where the response time is critical and the data are acquired in a stress situation which affects the quality of data significantly.     

A data mining approach considering missing values for the optimization of semiconductor-manufacturing processes

Due to the rapid development of information technologies, abundant data have become readily available. Data mining techniques have been used for process optimization in many manufacturing processes in automotive, LCD, semiconductor, and steel production, among others. However, a large amount of missing values occurs in the data set due to several causes (e.g., data discarded by gross measurement errors, measurement machine breakdown, routine maintenance, sampling inspection, and sensor failure), which frequently complicate the application of data mining to the data set. This study proposes a new procedure for optimizing processes called missing values-Patient Rule Induction Method (m-PRIM), which handles the missing-values problem systematically and yields considerable process improvement, even if a significant portion of the data set has missing values. A case study in a semiconductor manufacturing process is conducted to illustrate the proposed procedure.     

互联网时代商业空间活力缺失及多元回归路径探索——以西安骡马市商业步行街为例

互联网时代,电子商务高奏的凯歌,似乎成为实体商业的"挽歌",作为城市核心空间的商业空间在内外冲击下出现了不同程度的衰败。本文基于城市更新的背景,以活力为切入点,以西安骡马市商业街为例,通过人群密度、可达性、空间品质等多角度揭示活力缺失现象及原因,并从开发导向、业态、文化、空间、和机制等方面探索出切实可行的多元活力回归路径,进而引导商业空间转型和发展,以期恢复往日的活力。     

Blind separation of fetal electrocardiograms by annealed expectation maximization

This work explores blind source separation of fetal electrocardiograms by annealed expectation maximization (AEM). The AEM method improves the traditional EM method by relaxation under an annealing process, which is recruited to avoid trappings of tremendous spurious local minima within an objective function that inversely measures quantitative performance of a demixing structure. The derived objective function depends on the demixing structure as well as a set of membership vectors that represent missing data toward encoding statistical dependency of retrieved independent sources. Under the annealing process, the derived E and M steps are iteratively performed to search for expectations of membership vectors and minimizers of the objective function. The state number of membership vectors is related to modulate discretization of observations. Its effects on extraction of fetal electrocardiograms and reliability of the AEM method for blind source separation are extensively explored by numerical simulations.     

Logical operators of some statistical computing packages and missing values

When using statistical computer packages in general, we rely on the results they produce. We are aware that numerical approximations are made and trust that the best algorithms are chosen to do them. Most manuals give us instructions about precision of calculations and some report how missing values are administered. What we are unaware of is that some packages can invent results when creating atomic formulas and compounding complex formulas out of atomic ones, what inflates sample sizes, and possibly leads us to incorrect statistical decisions. Two simple indicator variables, with missing values positioned so the results are always missing values, were tested as numerical, as logical and as character variables, by compounding them through connective ‘and’ (&) and ‘or’ (|) to form new indicator variables. The results show that one of the three very known packages does not, statistically, correctly handle missing values, and the three make atomic formulas out of character variables assigning the value false (0) for missing value, what can be said an statistical error. The conclusion is that statisticians and users of statistics must be aware of the capabilities of logically operating missing values of the statistical packages they use, otherwise wrong statistical decisions can be made. And that programmers of statistical packages should correct their algorithms in order to not permit their packages invent non-existing values.     

缺少投影角度与时变物体断层重建的外推恢复正弦图方法

本文针对缺少投影角度及物体时变这两类缺损投影数据问题，在分析了限宽间区域及限频域图像的雷当变换的若干性质基础上，分别导出基于最小均方误差准则的外推公式，从而得到完整的正弦图的最佳估计，基于这一外推所得的正弦图，可以利用传统的断层重建方法对物体进行重建，本文给出了外推公式的推导过程有其求解的迭代算法，计算机模拟结果显示了这一技术的有效性。     

Covariance miss-specification and the local influence approach in sensitivity analyses of longitudinal data with drop-outs

Our work examines the performance of proposed local influence diagnostics applied to multivariate normal longitudinal data with drop-outs: these diagnostics prove to be ambiguous as they are sensitive not only to the presence of anomalous records, as intended, but also, unfortunately, to the misspecification of the longitudinal covariance structure of the response. We suggest an unambiguous index for detecting covariance misspecification, and recommend that an analyst use this index first to confirm that the covariance structure is well specified before attempting to interpret the influence diagnostics.     

SYMMETRIC STABLE SEQUENCES WITH MISSING OBSERVATIONS

Abstract. A procedure for evaluating optimal linear estimates of missing values in the minimum dispersion sense is proposed for stationary symmetric stable processes. Analytical expressions for the estimates are obtained for the autoregressive moving-average process and it is shown that the finite variance setting results are special cases. Cases of one and more than one missing value are considered.