Many attempts have been made to achieve successful recognition of handwritten digits. We report our results of using statistical method on handwritten digit recognition. A digitized handwritten numeral can be represented by an image with grayscales. The image includes features that are mapped into two-dimensional space with row and column coordinates. Based on this structure, two-dimensional penalized signal logistic regression (PSR) is applied to the recognition of handwritten digits. The data set is taken from the USPS zip code database that contains 7219 training images and 2007 test images. All the images have been deslanted and normalized into 16 x 16 pixels with various grayscales. The PSR method constructs a coefficient surface using a rich two-dimensional tensor product B-splines basis, so that the surface is more flexible than needed. We then penalize roughness of the coefficient surface with difference penalties on each coefficient associate with the rows and columns of the tensor product B-splines. The optimal penalty weight is found in several minutes of iterative operations. A competitive overall recognition error rate of 8.97% on the test data set was achieved. We will also review an artificial neural network approach for comparison. By using PSR, it requires neither long learning time nor large memory resources. Another advantage of the PSR method is that our results are obtained on the original USPS data set without any further image preprocessing. We also found that PSR algorithm was very capable to cope with high diversity and variation that were two major features of handwritten digits.

The ironwood tree (Casuarina equisetifolia), a protector of coastlines of the sub-tropical and tropical Western Pacific, is in decline on the island of Guam where aggressive data collection and efforts to mitigate the problem are underway. For each sampled tree the level of decline was measured on an ordinal scale consisting of five categories ranging from healthy to near dead. Several predictors were also measured including tree diameter, fire damage, typhoon damage, presence or absence of termites, presence or absence of basidiocarps, and various geographical or cultural factors. The five decline response levels can be viewed as categories of a multinomial distribution where the multinomial probability profile depends on the levels of these various predictors. Such data structure is well suited to a proportional odds model thereby leading to odds ratios involving cumulative probabilities which can be estimated and summarized using information from the predictor coefficient. Various modeling techniques were applied to address data set issues: reduced logistic models, spatial relationships of residuals using latitude and longitude coordinates, and correlation structure induced by the fact that trees were sampled in clusters at various sites. Among our findings, factors related to ironwood decline were found to be basidiocarps, termites, and level of human management.

Thesis (Ph. D.)--Northwestern University, 1977. / Vita. eContent provider-neutral record in process. Description based on print version record. Includes bibliographical references (leaves 224-231).

Thesis (Ph.D.)--University of Illinois at Urbana-Champaign, 2009. / Source: Dissertation Abstracts International, Volume: 70-06, Section: B, page: 3815. Adviser: Alejandro Lleras. Includes bibliographical references (leaves 78-84) Available on microfilm from Pro Quest Information and Learning.

Implicit and explicit memory for novel, three-dimensional possible and impossible objects was investigated using two memory tasks. The object decision task required subjects to determine whether line drawings depicted possible or impossible objects. Subjects who studied the global, three-dimensional properties of the possible objects showed superior performance for those items on a subsequent object decision test, while subjects who studied the local features of these same objects showed no improvement. The impossible figures showed no facilitation due to prior exposure. Recognition performance was unaffected by the type of encoding task. These results provided support for the hypothesis that processes mediating priming in the object decision task require access to a specific structural description formed in the initial viewing.

published_or_final_version / Statistics / Doctoral / Doctor of Philosophy

The influence of spatial attention location on figure-ground organization was assessed in an experiment in which a cued detection paradigm was paired with a figure-ground task. Viewers' fixation was held constant while a cued detection task directed their attention to a location in one of two regions within a figure-ground stimulus. Viewers were more likely to see a meaningless shape (low denotative region) as figure when spatial attention was allocated to that region than when it was allocated elsewhere. The location of spatial attention had no influence on whether or not a meaningful shape (high denotative region) was seen as figure. The results are discussed in terms of direct and indirect routes through which attention might influence figure-ground organization.

Two experiments were conducted to explore implicit memory within the auditory modality. It was hypothesized that there would be significant auditory priming in a masked word discrimination task. It was also hypothesized that subjects who performed an elaborative encoding task would show little or no advantage in the priming test when compared to subjects who performed a structural encoding task, but they would perform significantly better on a recognition test. It was also hypothesized that, with an appropriate encoding task, subjects would show a sensitivity to voice changes in the priming test. The results indicated significant auditory priming, and elaborative encoding had a greater effect on recognition performance than priming. No sensitivity to voice change was found. The findings are interpreted as support for the theory that auditory priming is mediated by a perceptual representation system.

Pain sensitivity is importantly affected by a number of factors, and may best be evaluated using a biopsychosocial framework. Although such a framework has been applied to the study and treatment of pain in humans, the concept of pain as a multidimensional experience in rodent models has been given considerably less attention. There is evidence, however, that pain is significantly modulated by a variety of psychosocial factors; therefore, identifying such factors is important, especially considering the extensive use of rodent subjects in pain research. / A previous study performed in our laboratory described a within-cage order-of-testing effect that suggested the social communication of pain amongst mice, a finding that inspired the research described in this dissertation. Using novel experimental paradigms, the research presented describes the effects of the immediate social environment on pain sensitivity in the laboratory mouse, as well as the converse, the effect of pain on social interactions among mice. Social conditions in pain experiments vary with respect to familiarity (i.e., cagemate versus stranger), sex, and pain state of the conspecific and such factors appear to significantly modulate both pain sensitivity and social behaviour. / Here I present social modulation of pain as evidence for empathy in mice, such that mice observing a cagemate in pain exhibit significant hypersensitivity to a noxious stimulus; such modulation is dependent on visual cues. I also show that varying perceived social threat by permitting or limiting full physical contact in stranger male dyads significantly modulates pain behaviour resulting in testosterone-dependent stress-induced analgesia or hyperalgesia, respectively. Furthermore, I show that females display heightened social approach toward a cagemate displaying pain behaviour, that this social approach is also observed in mice lacking the oxytocin receptor gene, and that such behaviour may have analgesic properties. Finally, in light of our initial finding that pain is visually communicated, I present a novel coding system for facial expressions of pain in the mouse. / In summary, these findings suggest the importance of accounting for psychosocial factors affecting pain sensitivity in the laboratory mouse, and suggest potential animal models for such complex social processes as empathy and prosociality. Furthermore, facial expression is a novel dependent measure of pain that may provide a more complete description of an animal's pain experience. / La sensibilité à la douleur est grandement influencée par un certain nombre de facteurs et pourrait être évaluée plus adéquatement selon un cadre bio-psycho-social. Bien qu'un tel cadre ait été appliqué à l'étude et au traitement de la douleur chez l'humain, le concept de douleur en tant qu'expérience multidimensionnelle chez le rongeur a bénéficié de beaucoup moins d'attention. Pourtant, il est démontré que la douleur est significativement modulée par un ensemble de facteurs psychosociaux; l'identification de ces facteurs demeure donc importante, particulièrement lorsqu'on considère l'usage considérable des rongeurs dans le domaine de la recherche sur la douleur. / D'après une étude réalisée précédemment par notre équipe, les souris qui cohabitent et qui sont témoins de leur souffrance mutuelle sont plus sensibles à la douleur que les souris soumises au test à la douleur de manière individuelle. Ce constat de l'influence de la cohabitation et de l'ordre selon lequel les souris sont testées, évoquant l'existence d'une communication sociale, a inspiré les recherches décrites dans cette thèse. Utilisant de nouveaux paradigmes expérimentaux, nos travaux décrivent les effets de l'environnement social immédiat sur la sensibilité à la douleur chez la souris de laboratoire et, inversement, les effets de cette douleur sur ses interactions sociales. Les conditions sociales varient en fonction du degré de familiarité (c.-à-d. souris consoeurs contre souris étrangères), du sexe et du niveau de douleur de la congénère, tous des facteurs qui semblent moduler de manière significative la sensibilité à la douleur et le comportement social. / J'expose ici que la modulation sociale de la douleur constitue une preuve d'empathie chez la souris de sorte que des souris qui cohabitent et qui voient leurs consoeurs être en proie à la douleur affichent une hypersensibilité significative lors des tests de nociception; une telle modulation dépend de signaux visuels. Je démontre également que le changement de perception de menace sociale parmi une paire de souris mâles étrangères affecte leur sensibilité à la douleur, entraînant les réponses dépendantes à la testostérone et induites par le stress que sont l'analgésie, obtenue lorsqu'on permet un contact physique complet ou l'hyperalgésie, obtenue lorsqu'un tel contact est limité. Par ailleurs, je montre que les femelles manifestent une approche sociale plus marquée envers une consoeur en proie à la douleur, que cette approche se voit aussi chez des souris dépourvues du gène codant pour le récepteur oxytocine et qu'un tel comportement pourrait avoir des propriétés analgésiques. Enfin, à la lumière de notre découverte initiale révélant que la douleur est communiquée de façon visuelle, je présente un système de codage inédit des expressions faciales de douleur chez la souris. / En résumé, ces résultats suggèrent l'importance de tenir compte des facteurs psychosociaux dans l'analyse des facteurs pouvant influencer la sensibilité à la douleur chez la souris de laboratoire et proposent des modèles animaux potentiels pour l'étude de processus sociaux complexes comme l'empathie et la prosocialité. Par ailleurs, l'expression faciale constitue une mesure dépendante originale susceptible de décrire de manière plus complète l'expérience de douleur chez l'animal.

No description available.