Measurements of generalisation based on information geometry

Neural networks are statistical models and learning rules are estimators. In this paper a theory for measuring generalisation is developed by combining Bayesian decision theory with information geometry. The performance of an estimator is measured by the information divergence between the true distribution and the estimate, averaged over the Bayesian posterior. This unifies the majority of error measures currently in use. The optimal estimators also reveal some intricate interrelationships among information geometry, Banach spaces and sufficient statistics.

Examination of the predictive value of structural magnetic resonance scans in bipolar disorder:a pattern classification approach

Background - Bipolar disorder (BD) is one of the leading causes of disability worldwide. Patients are further disadvantaged by delays in accurate diagnosis ranging between 5 and 10 years. We applied Gaussian process classifiers (GPCs) to structural magnetic resonance imaging (sMRI) data to evaluate the feasibility of using pattern recognition techniques for the diagnostic classification of patients with BD. Method - GPCs were applied to gray (GM) and white matter (WM) sMRI data derived from two independent samples of patients with BD (cohort 1: n = 26; cohort 2: n = 14). Within each cohort patients were matched on age, sex and IQ to an equal number of healthy controls. Results - The diagnostic accuracy of the GPC for GM was 73% in cohort 1 and 72% in cohort 2; the sensitivity and specificity of the GM classification were respectively 69% and 77% in cohort 1 and 64% and 99% in cohort 2. The diagnostic accuracy of the GPC for WM was 69% in cohort 1 and 78% in cohort 2; the sensitivity and specificity of the WM classification were both 69% in cohort 1 and 71% and 86% respectively in cohort 2. In both samples, GM and WM clusters discriminating between patients and controls were localized within cortical and subcortical structures implicated in BD. Conclusions - Our results demonstrate the predictive value of neuroanatomical data in discriminating patients with BD from healthy individuals. The overlap between discriminative networks and regions implicated in the pathophysiology of BD supports the biological plausibility of the classifiers.

Boolean Rings that are Baire Spaces

∗ The present article was originally submitted for the second volume of Murcia Seminar on Functional Analysis (1989). Unfortunately it has been not possible to continue with Murcia Seminar publication anymore. For historical reasons the present vesion correspond with the original one.

Subdifferentials of Performance Functions and Calculus of Coderivatives of Set-Valued Mappings

The paper contains calculus rules for coderivatives of compositions, sums and intersections of set-valued mappings. The types of coderivatives considered correspond to Dini-Hadamard and limiting Dini-Hadamard subdifferentials in Gˆateaux differentiable spaces, Fréchet and limiting Fréchet subdifferentials in Asplund spaces and approximate subdifferentials in arbitrary Banach spaces. The key element of the unified approach to obtaining various calculus rules for various types of derivatives presented in the paper are simple formulas for subdifferentials of marginal, or performance functions.

Uniform Convergence of the Newton Method for Aubin Continuous Maps

* This work was supported by National Science Foundation grant DMS 9404431.

Variational Principles for Monotone and Maximal Bifunctions

2000 Mathematics Subject Classification: 49J40, 49J35, 58E30, 47H05

Upper and Lower Bounds in Relator Spaces

2000 Mathematics Subject Classification: 06A06, 54E15

Weak Compatibility and Common Fixed Point Theorems for A-Contractive and E-expansive Maps in Uniform Spaces

2000 Mathematics Subject Classification: 47H10, 54E15.

A Note On Quasi-Lindelöf Spaces

Петра Г. Стайнова - Квази-линдельофовите пространства са въведени от Архангелски като усилване на слабо-линдельофовите. В тази статия се разглеждат няколко свойства на квази-линдельофовите пространства и се правят сравнения със съответни ре- зултати за линдельофовите и слабо-линдельофовите пространства. Дадени са няколко примера, включително на слабо-линдельофово пространство, което не е квази-линдельофово; на пространство, което е топологично произведение на две линдельофови, но не е дори квази-линдельофово, и на пространство, което е квази-линдельофово, но не Суслиново. Накрая са поставени няколко отворени въпроси.

Little G. T. for lp-lattice summing operators

2000 Mathematics Subject Classification: 46B28, 47D15.

Moduli stacks of polarized K3 surfaces in mixed characteristic

2000 Mathematics Subject Classification: 14J28, 14D22.

Weakly Compact Generating and Shrinking Markusevic Bases

2000 Mathematics Subject Classification: 46B30, 46B03.

Multidimensional classification and diagnosis of leukemia in flow cytometry

This research is to establish new optimization methods for pattern recognition and classification of different white blood cells in actual patient data to enhance the process of diagnosis. Beckman-Coulter Corporation supplied flow cytometry data of numerous patients that are used as training sets to exploit the different physiological characteristics of the different samples provided. The methods of Support Vector Machines (SVM) and Artificial Neural Networks (ANN) were used as promising pattern classification techniques to identify different white blood cell samples and provide information to medical doctors in the form of diagnostic references for the specific disease states, leukemia. The obtained results prove that when a neural network classifier is well configured and trained with cross-validation, it can perform better than support vector classifiers alone for this type of data. Furthermore, a new unsupervised learning algorithm---Density based Adaptive Window Clustering algorithm (DAWC) was designed to process large volumes of data for finding location of high data cluster in real-time. It reduces the computational load to ∼O(N) number of computations, and thus making the algorithm more attractive and faster than current hierarchical algorithms.

Spaces of Order: An African Poetics of Space

“Spaces of Order” argues that the African novel should be studied as a revolutionary form characterized by aesthetic innovations that are not comprehensible in terms of the novel’s European archive of forms. It does this by mapping an African spatial order that undermines the spatial problematic at the formal and ideological core of the novel—the split between a private, subjective interior, and an abstract, impersonal outside. The project opens with an examination of spatial fragmentation as figured in the “endless forest” of Amos Tutuola’s The Palmwine Drinkard (1952). The second chapter studies Chinua Achebe’s Things Fall Apart (1958) as a fictional world built around a peculiar category of space, the “evil forest,” which constitutes an African principle of order and modality of power. Chapter three returns to Tutuola via Ben Okri’s The Famished Road (1991) and shows how the dispersal of fragmentary spaces of exclusion and terror within the colonial African city helps us conceive of political imaginaries outside the nation and other forms of liberal political communities. The fourth chapter shows Nnedi Okorafor—in her 2014 science-fiction novel Lagoon—rewriting Things Fall Apart as an alien-encounter narrative in which Africa is center-stage of a planetary, multi-species drama. Spaces of Order is a study of the African novel as a new logic of world making altogether.

Governmentality and exclusion in post-disaster spaces : conducting the conduct of the survivors of Typhoon Sendong in Cagayan de Oro, Philippines

Lorsque les aléas naturels se déroulent en catastrophes, les réponses des religieux, de l’Etat, et d’autres acteurs puissants dans une société révèlent à la fois les relations complexes entre ces parties et leur pouvoir dans la production des espaces auxquelles les survivants accèdent. La réponse en cas de catastrophe comprend la création d’espaces post-catastrophes, tels que des centres d’évacuation, des logements de transition et des sites de réinstallation permanente, qui ciblent spécifiquement un sous-ensemble particulier de survivants, et visent à les aider à survivre, à faire face, et à se remettre de la catastrophe. Les acteurs puissants dans une société dirigent les processus de secours, de récupération et de reconstruction sont des acteurs puissants qui cherchent à problématiser et à rendre un problème technique dans des termes qu’ils sont idéalement placés pour aborder à travers une variété d'interventions. Ce projet de recherche vise à répondre à la question: où les survivants d'une catastrophe reconstruisent-ils leurs vies et leurs moyens de subsistance? Il enquête sur un cas spécifique de la migration environnementale dans laquelle des dizaines de milliers d'habitants ont été déplacés de façon permanente et temporaire de leurs résidences habituelles après le typhon Sendong à Cagayan de Oro, Philippines en 2011. La recherche est basée sur des entretiens avec les acteurs puissants et les survivants, des vidéos participatives réalisées par des survivants pauvres urbains, et des activités de cartographie. L’étude se fonde sur la théorie féministe, les études de migration, les études dans la gouvernementalité, la recherche sur les changements de l’environnement planétaire, et les études régionales afin de situer les diverses expériences de la migration dans un contexte géographique et historique. Cette thèse propose une topographie critique dans laquelle les processus et les pratiques de production d’espaces post-catastrophe sont exposés. Parce que l’espace est nécessairement malléable, fluide, et relationnelle en raison de l'évolution constante des activités, des conflits, et des expériences qui se déroulent dans le paysage, une analyse de l'espace doit être formulée en termes de relations sociales qui se produisent dans et au-delà de ses frontières poreuses. En conséquence, cette étude explore comment les relations sociales entre les survivants et les acteurs puissants sont liées à l’exclusion, la gouvernementalité, la mobilité, et la production des espaces, des lieux et des territoires. Il constate que, si les trajectoires de migration de la plupart des survivants ont été confinés à l'intérieur des limites de la ville, les expériences de ces survivants et leur utilisation des espaces urbains sont très différentes. Ces différences peuvent être expliquées par des structures politiques, économiques, et sociales, et par les différences religieuses, économiques, et de genre. En outre, il fait valoir que les espaces post-catastrophe doivent être considérés comme des «espaces d’exclusion» où les fiduciaires exercent une rationalité gouvernementale. C’est-à-dire, les espaces post-catastrophe prétendument inclusives servent à marginaliser davantage les populations vulnérables. Ces espaces offrent aussi des occasions pour les acteurs puissants dans la société philippine d'effectuer des interventions gouvernementales dans lesquelles certaines personnes et les paysages sont simplifiées, rendues lisibles, et améliorés.