Tesis sobre el tema "Reweighting"

In this thesis, we focus on the application of covariate reweighting with Lasso-style methods for regression in high dimensions, particularly where p ≥ n. We apply a particular focus to the case of sparse regression under a-priori grouping structures. In such problems, even in the linear case, accurate estimation is difficult. Various authors have suggested ideas such as the Group Lasso and the Sparse Group Lasso, based on convex penalties, or alternatively methods like the Group Bridge, which rely on convergence under repetition to some local minimum of a concave penalised likelihood. We propose in this thesis a methodology that uses concave penalties to inspire a procedure whereupon we compute weights from an initial estimate, and then do a single second reweighted Lasso. This procedure -- the Co-adaptive Lasso -- obtains excellent results in empirical experiments, and we present some theoretical prediction and estimation error bounds. Further, several extensions and variants of the procedure are discussed and studied. In particular, we propose a Lasso style method of doing additive isotonic regression in high dimensions, the Liso algorithm, and enhance it using the Co-adaptive methodology. We also propose a method of producing rules based regression estimates for high dimensional non-parametric regression, that often outperforms the current leading method, the RuleFit algorithm. We also discuss extensions involving robust statistics applied to weight computation, repeating the algorithm, and online computation.

Noisy labels could cause severe degradation to the classification performance. Especially for deep neural networks, noisy labels can be memorized and lead to poor generalization. Recently label noise robust deep learning has outperformed traditional shallow learning approaches in handling complex input data without prior knowledge of label noise generation. Learning from noisy labels by importance reweighting is well-studied. Existing work in this line using deep learning failed to provide reasonable importance reweighting criterion and thus got undesirable experimental performances. Targeting this knowledge gap and inspired by domain adaptation, we propose a novel label noise robust deep learning approach by importance reweighting. Noisy labeled training examples are weighted by minimizing the maximum mean discrepancy between the loss distributions of noisy labeled and clean labeled data. In experiments, the proposed approach outperforms other baselines. Results show a vast research potential of applying domain adaptation in label noise problem by bridging the two areas. Moreover, the proposed approach potentially motivate other interesting problems in domain adaptation by enabling importance reweighting to be used in deep learning.
Felaktiga annoteringar kan sänka klassificeringsprestanda.Speciellt för djupa nätverk kan detta leda till dålig generalisering. Nyligen har brusrobust djup inlärning överträffat andra inlärningsmetoder när det gäller hantering av komplexa indata Befintligta resultat från djup inlärning kan dock inte tillhandahålla rimliga viktomfördelningskriterier. För att hantera detta kunskapsgap och inspirerat av domänanpassning föreslår vi en ny robust djup inlärningsmetod som använder omviktning. Omviktningen görs genom att minimera den maximala medelavvikelsen mellan förlustfördelningen av felmärkta och korrekt märkta data. I experiment slår den föreslagna metoden andra metoder. Resultaten visar en stor forskningspotential för att tillämpa domänanpassning. Dessutom motiverar den föreslagna metoden undersökningar av andra intressanta problem inom domänanpassning genom att möjliggöra smarta omviktningar.

In our everyday lives, we must often weigh the different attributes of items in order to select the item that best fits our current goals, allowing us to make optimal decisions. Construal Level Theory proposes a psychological mechanism for re-weighting attributes, utilizing selective attention as the process by which we implement self-control. It has been hypothesized that switching attention between attributes is facilitated by the suppression of cortical oscillations over posterior brain regions within the alpha (8-12 Hz) frequency range. To test this idea, we re-examined previously collected whole-brain electroencephalography (EEG) data from a dietary choice experiment in which participants made decisions naturally or with a weight loss incentive. Prior analysis found that although hungry subjects primarily relied on taste properties while responding naturally, they increased their behavioral and neural weighting of health when motivated to lose weight. Reanalyzing this data using time-frequency analyses, we compared alpha oscillations related to healthy versus unhealthy foods under natural and self-control conditions. We predicted that when participants exercised self-control we would see suppression of alpha oscillations over occipital sensors starting around 400 ms post-stimulus onset, for trials presenting healthy relative to unhealthy foods; no such suppression should appear during natural responding when ignoring health information. Consistent with our hypothesis, we found a significant decrease in alpha oscillations over occipital sensors between 440 and 800 ms post-stimulus onset for healthy compared to unhealthy items in the self-control condition. No such effect was seen for health information in natural choice, or for taste. Our findings extend previous research by linking alpha band suppression to the neural re-weighting of multiple attributes, suggesting a neuro-cognitive mechanism for self-control that uses selective attention to choose between multiple attributes.

Context: Individuals with chronic ankle instability (CAI) rely more on visual information during postural control due to impaired proprioceptive function. The increased reliance on visual information may increase the risk of injury when their vision is limited during complex sports activities. Stroboscopic glasses may help elicit sensory reweighting during postural control. Therefore, we assumed that the glasses would induce and train CAI patients to reweight sensory information for the somatosensory system during dynamic balance training. Purpose: (1) to identify the effects of the 4-week dynamic balance training on the reliance of visual information during postural control in patients with CAI and (2) to compare the effects of the 4-week dynamic balance with and without stroboscopic glasses on postural control in patients with CAI. Methods: This study was a randomized controlled trial. Twenty-eight CAI patients were equally assigned to one of 2 groups: a strobe group (6 males and 8 females) or a control group (8 males and 6 females). The 4-week dynamic balance training consisted of multiple single-legged exercises. The strobe group wore stroboscopic glasses during the training, but the control group did not. The main outcome measures included the following: self-reported function measures, static postural control (center of posture (COP)-based measures), and dynamic postural control including the Dynamic Postural Stability Index (DPSI), and the Star Excursion Balance Test (SEBT). There were 3 visual conditions in the static postural control (eyes-open (EO), strobe vision (SV), and eyes-closed (EC)), and 2 conditions in the dynamic postural control (EO and SV). Two-way randomized block ANOVAs were used to assess changes in postural control in each group and condition by using pretest-posttest mean differences. Results: The strobe group showed a higher difference in COP velocity in medial-lateral direction (VelML) and vertical stability index (VSI) under the SV condition compared with the control group (p = .005 and .004, respectively). In addition, the strobe group had significant decreases in VelML, DPSI, and VSI at the posttest compared with the pretest (p = .0001, .01, and .005, respectively). Conclusion: The 4-week dynamic balance training with stroboscopic glasses appeared to be effective in improving postural control and altering visual reliance in patients with CAI.

When using visual and proprioceptive information to plan a reach, it has been proposed that the brain combines these cues to estimate the object and/or limb’s location. Specifically, according to the maximum-likelihood estimation (MLE) model, more reliable sensory inputs are assigned a greater weight (Ernst & Banks, 2002). In this research we examined if the brain is able to adjust which sensory cue it weights the most. Specifically, we asked if the brain changes how it weights sensory information when the availability of a visual cue is manipulated. Twenty-four healthy subjects reached to visual (V), proprioceptive (P), or visual + proprioceptive (VP) targets under different visual delay conditions (e.g. on V and VP trials, the visual target was available for the entire reach, it was removed with the go-signal or it was removed 1, 2 or 5 seconds before the go-signal). Subjects completed 5 blocks of trials, with 90 trials per block. For 12 subjects, the visual delay was kept consistent within a block of trials, while for the other 12 subjects, different visual delays were intermixed within a block of trials. To establish which sensory cue subjects weighted the most, we compared endpoint positions achieved on V and P reaches to VP reaches. Results indicated that all subjects weighted sensory cues in accordance with the MLE model across all delay conditions and that these weights were similar regardless of the visual delay. Moreover, while errors increased with longer visual delays, there was no change in reaching variance. Thus, manipulating the visual environment was not enough to change subjects’ weighting strategy, further i

Orientador: José Angelo Barela
Banca: Sérgio Tosi Rodrigues
Banca: Ronald Dennis Paul Kenneth Clive Ranvaud
Banca: Renato de Moraes
Banca: Dora S. Fix Ventura
Resumo: Três experimentos foram propostos para investigar os ajustes dinâmicos nos pesos de múltiplas fontes de informação sensorial. O primeiro experimento investigou as respostas posturais de 18 bebês frente às mudanças abruptas na amplitude do estímulo visual. Eles permaneceram sentados dentro de uma sala móvel por 8 tentativas de 60 segundos cada. A sala ficou estacionária na primeira tentativa. Nas sete tentativas seguintes, a sala foi oscilada em 0,2 Hz com amplitude de 1,1 cm, com exceção da quinta tentativa, na qual a sala foi movimentada em amplitude mais alta (3,2 cm). Os resultados mostraram fraco acoplamento entre informação visual e oscilação corporal. Contudo, a variabilidade de oscilação foi maior em bebês experientes no sentar independente. Nós concluímos que bebês não foram capazes de se adaptar às pequenas alterações na amplitude do estímulo visual. O segundo experimento investigou como o controle postural de crianças se adapta às abruptas mudanças no ambiente visual. Trinta crianças de 4, 8 e 12 anos de idade e dez adultos, permaneceram em pé dentro de uma sala móvel. A situação experimental foi similar a do primeiro experimento exceto que a amplitude baixa da sala foi de 0,5 cm e a amplitude alta foi de 3,2 cm. As respostas posturais de crianças mais velhas e adultas diminuíram mais para o estímulo visual do que aquelas de crianças mais novas quando a amplitude da sala foi aumentada. A variabilidade de oscilação diminuiu com a idade e foi maior durante a tentativa de alta amplitude. Crianças tão novas quanto 4 anos de idade já têm desenvolvida a capacidade de rapidamente diminuir a influência do estimulo visual. Contudo, os mais altos valores de ganho e variabilidade residual para crianças de 4 e 8 anos de idade sugerem que elas não apresentam respostas totalmente calibradas ao nível adulto... (Resumo completo, clicar acesso eletrônico abaixo)
Abstract: Three experiments were designed to investigate the dynamic adjustments in the weights of multiple sensory modalities. The first experiment investigated the postural responses of 18 infants to abrupt changes in the amplitude of visual stimulus. They seated inside of a moving room for eight trials of 60 sec apiece. The room was stationary in the first trial. In the following seven trials, the room oscillated at 0.2 Hz with amplitude of 1.1 cm, with the exception of the fifth trial, in which the room moved at higher amplitude (3.2 cm). The results showed weak coupling between visual stimulus and body sway. However, sway variability of experienced sitters was higher in the high-amplitude trial. We concluded that infants were not able to adapt to low range of visual stimulus amplitude. The second experiment investigated how children’s postural control adapt to abrupt changes in the surrounding visual environment. Thirty children from 4-, 8- and 12-year olds and ten adults stood upright inside of a moving room. The experimental set-up was similar to the first experiment except that low-amplitude was 0.5 cm and high-amplitude was 3.2 cm. Body sway responses of old children and adults downweight more to the visual stimulus than young children when the amplitude of the room increased. Sway variability decreased with age and was largest during the high-amplitude trial. Children as young as four years of age have already developed the adaptive capability to quickly downweight visual information. However, the higher gain values and residual variability observed for the 4 and 8 year-old children suggest that they have not fully calibrated their response to the adult level. The third experiment investigated the postural responses of two sensory modalities measured simultaneously... (Complete abstract click electronic access below)
Doutor

Made available in DSpace on 2014-06-11T19:30:54Z (GMT). No. of bitstreams: 0 Previous issue date: 2007-08-06Bitstream added on 2014-06-13T19:19:36Z : No. of bitstreams: 1 zago_pfp_dr_rcla.pdf: 1617834 bytes, checksum: 193398c19868a245f447705bc0bf0861 (MD5)
Três experimentos foram propostos para investigar os ajustes dinâmicos nos pesos de múltiplas fontes de informação sensorial. O primeiro experimento investigou as respostas posturais de 18 bebês frente às mudanças abruptas na amplitude do estímulo visual. Eles permaneceram sentados dentro de uma sala móvel por 8 tentativas de 60 segundos cada. A sala ficou estacionária na primeira tentativa. Nas sete tentativas seguintes, a sala foi oscilada em 0,2 Hz com amplitude de 1,1 cm, com exceção da quinta tentativa, na qual a sala foi movimentada em amplitude mais alta (3,2 cm). Os resultados mostraram fraco acoplamento entre informação visual e oscilação corporal. Contudo, a variabilidade de oscilação foi maior em bebês experientes no sentar independente. Nós concluímos que bebês não foram capazes de se adaptar às pequenas alterações na amplitude do estímulo visual. O segundo experimento investigou como o controle postural de crianças se adapta às abruptas mudanças no ambiente visual. Trinta crianças de 4, 8 e 12 anos de idade e dez adultos, permaneceram em pé dentro de uma sala móvel. A situação experimental foi similar a do primeiro experimento exceto que a amplitude baixa da sala foi de 0,5 cm e a amplitude alta foi de 3,2 cm. As respostas posturais de crianças mais velhas e adultas diminuíram mais para o estímulo visual do que aquelas de crianças mais novas quando a amplitude da sala foi aumentada. A variabilidade de oscilação diminuiu com a idade e foi maior durante a tentativa de alta amplitude. Crianças tão novas quanto 4 anos de idade já têm desenvolvida a capacidade de rapidamente diminuir a influência do estimulo visual. Contudo, os mais altos valores de ganho e variabilidade residual para crianças de 4 e 8 anos de idade sugerem que elas não apresentam respostas totalmente calibradas ao nível adulto...
Three experiments were designed to investigate the dynamic adjustments in the weights of multiple sensory modalities. The first experiment investigated the postural responses of 18 infants to abrupt changes in the amplitude of visual stimulus. They seated inside of a moving room for eight trials of 60 sec apiece. The room was stationary in the first trial. In the following seven trials, the room oscillated at 0.2 Hz with amplitude of 1.1 cm, with the exception of the fifth trial, in which the room moved at higher amplitude (3.2 cm). The results showed weak coupling between visual stimulus and body sway. However, sway variability of experienced sitters was higher in the high-amplitude trial. We concluded that infants were not able to adapt to low range of visual stimulus amplitude. The second experiment investigated how children s postural control adapt to abrupt changes in the surrounding visual environment. Thirty children from 4-, 8- and 12-year olds and ten adults stood upright inside of a moving room. The experimental set-up was similar to the first experiment except that low-amplitude was 0.5 cm and high-amplitude was 3.2 cm. Body sway responses of old children and adults downweight more to the visual stimulus than young children when the amplitude of the room increased. Sway variability decreased with age and was largest during the high-amplitude trial. Children as young as four years of age have already developed the adaptive capability to quickly downweight visual information. However, the higher gain values and residual variability observed for the 4 and 8 year-old children suggest that they have not fully calibrated their response to the adult level. The third experiment investigated the postural responses of two sensory modalities measured simultaneously... (Complete abstract click electronic access below)

La simulación molecular presenta la ventaja de ofrecer un marco teórico importante para predecir propiedades termodinámicas y de transporte de fluidos con aplicaciones industriales. En este trabajo, se explotó está ventaja para predecir el equilibrio líquido vapor de agua, compuestos aromáticos y sus mezclas a condiciones tanto sub- como supercríticas. Se realizó una comparación de diferentes potenciales intermoleculares conocidos mediante el cálculo de propiedades termodinámicas de sistemas puros que sirvió de punto de partida para llevar a cabo una optimización de parámetros transferibles para un potencial intermolecular de agua y compuestos aromáticos. En el caso de agua, se llevo a cabo el análisis y evaluación de propiedades termodinámicas de un modelo simple de agua. En este modelo, las contribuciones electrostáticas se aproximan mediante interacciones de corto alcance en vez de las típicas fuerzas de Coulomb de largo alcance. En general, se han encontrado grandes desviaciones con respecto a los datos experimentales, tal como un valor de temperatura crítica de 360K, valor 50% alejado del valor experimental. Debido a que estos resultados nos indican la importancia de incluir las fuerzas de Coulomb en el modelo molecular empleado para reproducir correctamente las propiedades de agua, el trabajo de investigación se ha enfocado en la optimización de los parámetros de los potenciales TIP4P y SPC/E. Los resultados obtenidos muestran que es posible encontrar una mejor aproximación al punto crítico experimental a partir de la optimización del modelo SPC/E. Sin embargo, el buen acuerdo con los experimentos del modelo original a condiciones ambiente se pierde usando los parámetros del modelo optimizado. Por otro lado, la estimación de propiedades de compuestos aromáticos esta de acuerdo con los datos experimentales permitiendo la reproducción de la densidad de líquido saturado, presión de saturación y entalpía de vaporización para compuestos puros. Finalmente, en el caso de mezclas se ha aplicado el conjunto de parámetros obtenidos para aromáticos. Las propiedades termodinámicas de la mezcla binaria aromático-aromático y aromático agua son analizadas en un amplio rango de temperaturas y presiones. Las desviaciones encontradas entre los valores calculados y los experimentales sugieren aplicar un mejor método de optimización para sistemas puros o por otro lado, promover un potencial de interacción intermolecular más sofisticado. Las estimaciones a condiciones cercanas al punto crítico están en buen acuerdo con los datos experimentales.
La simulació molecular presenta l'avantatge d'oferir un marc teòric important per a cercar propietats termodinàmiques i de transport de fluids amb aplicacions industrials. En aquest treball es va explotar aquesta avantatge per predir l'equilibri líquid vapor d'aigua, components aromàtics i les seves mescles, tant a condicions sub com supercrítiques. Es va realitzar una comparació de diferents potencials intermoleculars, coneguts mitjançant el càlcul de propietats termodinàmiques de sistemes purs, que ha servit de punt de sortida per portar a terme una optimització de paràmetres transferibles per a un potencial intermolecular de propietats termodinàmiques d'un model simple d'aigua. En aquest model, les contribucions electrostàtiques s'aproximen mitjançant interaccions de curt abast en lloc de les típiques forces de Coulomb de llarg abast. En general, s'ha trobat grans desviacions respecte a les dades experimentals, tal com un valor de temperatura crítica de 360K, valor 50% allunyat del valor experimental. Degut a que aquests resultats ens indiquen la importància d'incloure les forces de Coulomb en el model molecular emprat per reproduir correctament les propietats d'aigua, el treball d'investigació s'ha enfocat en l'optimització dels paràmetres dels potencials TIP4P i SPC/E. Els resultats obtinguts mostren que és possible trobar una millor aproximació al punt crític experimental a partir de l'optimització del model SPC/E. No obstant, el bon acord amb els experiments del model original a condicions ambientals es perden usant els paràmetres del model optimitzat. Per altre banda, l'estimació de propietats de compostos aromàtics esta d'acord amb les dades experimentals permetent la reproducció de la densitat de líquid saturat, pressió de saturació i entalpia de vaporització per a compostos purs mitjançant el potencial AUA-Aromátics proposat. Finalment, en el cas de mescles s'ha aplicat el conjunt de paràmetres obtinguts per aromàtics. Les propietats termodinàmiques de la mescla binària aromàtic-aromàtic i aromàtic-aigua són analitzades en un ample rang de temperatures i pressions. Les desviacions trobades entre els valors calculats i els experimentals suggereixen aplicar un millor mètode d'optimització per a sistemes purs o, per altre banda, promoure un potencial d'interacció intermolecular més sofisticat. Les estimacions en condicions properes al punt crític tenen un bon acord amb les dades experimentals.
Molecular simulation presents the advantage of providing a unified theoretical framework to model fluid properties for industrial applications. In this work we exploit this advantage to predict thermodynamic properties of pure water and aromatics and their mixtures at sub- and supercritical conditions. A comprehensive comparison of different intermolecular potentials has been carried out in order to analyze model predictions for pure component properties. In addition, an optimization of transferable parameters has been performed for an intermolecular potential for aromatics and water. In the case of water, an analysis and evaluation of the thermodynamic properties of a simple model has been performed. In this model, the electrostatic contributions are approximated by short-range interactions instead of the typical long-range Coulombic forces. On the whole, we found huge deviations with experimental data, such as a critical temperature value of 360K, 50% far away from the experimental value. Since, these calculations indicate the importance of including the electrostatic contribution in order to correctly model water, we also focus on reproducing critical properties from an optimization of the well known TIP4P and SPC/E water model parameters. Results obtained show that a better approximation to the critical point prediction is possible from the optimization of the SPC/E parameters, however, the good agreement with experiments for the original model at room conditions vanishes using the optimized parameters. On the other hand, thermodynamic property estimations of aromatic molecules are in good agreement with experimental data and we are able to reproduce saturation liquid densities, saturation pressures, vaporization enthalpies and liquid structure for pure compounds. Finally, in the case of mixtures, we applied the optimized set of parameters obtained for aromatics. The thermodynamic properties of binary aromatic-aromatic and aromatic water mixtures are analyzed over a wide range of temperatures and pressures. Deviations between the predicted and experimental values are found at low temperatures and high densities suggesting that a better optimization process needs to be performed for the pure systems or a more sophisticated intermolecular interaction potential is needed. Nevertheless, the estimations close to critical conditions are in good agreement with experimental data.

The thesis apply to reliability problems approach of Voronoi tessellation, typically used in the field of samples designs evaluation and for Monte Carlo samples reweighing. It is shown, this general technique estimation converges to that of Importance Sampling method despite it does not rely on Importance Sampling's auxiliary density. Consequently, reliability analysis could be divided into sampling itself and assessment of simulation results. As an extension of this idea, adaptive statistical sampling using QHull library was attempted.

Le 1er chapitre présente une série de 50 ans sur les hauts revenus chiliens basée sur des données fiscales et comptes nationaux. L’étude contredit les enquêtes, selon lesquelles les inégalités diminuent les 25 dernières années. Au contraire, elles changent de direction à partir de 2000. Le Chili est parmi les pays les plus inégalitaires de l’OCDE et l’Amérique latine. Le 2ème chapitre mesure la sous-estimation des revenus factoriels dans les données distributives. Les ménages ne reçoivent que 50% des revenus du capital brut, par opposition aux firmes. L’hétérogénéité des taux de réponse et autres problèmes font que les enquêtes ne capturent que 20% de ceux-ci, contre 70% du revenu du travail. Cela sous-estime l’inégalité,dont les estimations deviennent insensibles à la "capital share" et sa distribution. Je formalise à partir d’identités comptables pour ensuite calculer des effets marginaux et contributions aux variations d’inégalité. Le 3ème chapitre présente une méthode pour ajuster les enquêtes. Celles-ci capturent souvent mal le sommet de la distribution. La méthode présente plusieurs avantages par rapport aux options précédentes : elle est compatible avec les méthodes de calibration standard ; elle a des fondements probabilistes explicites et préserve la continuité des fonctions de densité ; elle offre une option pour surmonter les limites des supports d’enquête bornées; et elle préserve la structure de micro données en préservant la représentativité des variables sociodémographiques. Notre procédure est illustrée par des applications dans cinq pays, couvrant à la fois des contextes développés et moins développés
The 1st chapter presents historical series of Chilean top income shares over a period of half a century, mostly using data from tax statistics and national accounts. The study contradicts evidence based on survey data, according to which inequality has fallen constantly over the past 25 years. Rather, it changes direction, increasing from around the year 2000. Chile ranks as one of the most unequal countries among both OECD and Latin American countries over the whole period of study. The 2nd chapter measures the underestimation of factor income in distributive data. I find that households receive only half of national gross capital income,as opposed to corporations. Due to heterogeneous non-response and misreporting, Surveys only capture 20% of it, vs. 70% of labor income. This understates inequality estimates, which become insensitive to the capital share and its distribution. I formalize this system based on accounting identities. I then compute marginal effects and contributions to changes in fractile shares. The 3rd chapter, presents a method to adjust surveys. These generally fail to capturethe top of the income distribution. It has several advantages over previous ones: it is consistent with standard survey calibration methods; it has explicit probabilistic foundations and preserves the continuity of density functions; it provides an option to overcome the limitations of bounded survey-supports; and it preserves the microdata structure of the survey

La non-réponse, rencontrée dans la plupart des enquêtes épidémiologiques, est génératrice de biais de sélection (qui, dans ce cas est un biais de non-réponse) lorsqu’elle est liée aux variables d’intérêt. En surveillance épidémiologique, dont un des objectifs est d’estimer des prévalences, on a souvent recours à des enquêtes par sondage. On est alors confronté à la non-réponse totale et on peut utiliser des méthodes issues de la statistique d’enquête pour la corriger. Le biais de non-réponse peut être exprimé comme le produit de l’inverse du taux de réponse et de la covariance entre la probabilité de réponse et la variable d’intérêt. Ainsi, deux types de solution peuvent généralement être envisagés pour diminuer ce biais. La première consiste à chercher à augmenter le taux de réponse au moment de la planification de l’enquête. Cependant, la maximisation du taux de réponse peut entraîner d’autres types de biais, comme des biais de mesure. Dans la seconde, après avoir recueilli les données, on utilise des informations liées a priori aux variables d’intérêt et à la probabilité de réponse, et disponibles à la fois pour les répondants et les non-répondants pour calculer des facteurs correctifs. Cette solution nécessite donc de disposer d’informations sur l'ensemble de l'échantillon tiré au sort (que les personnes aient répondu ou non) ; or ces informations sont en général peu nombreuses. Les possibilités récentes d'accès aux bases médico-administratives (notamment celles de l'assurance maladie) ouvrent de nouvelles perspectives sur cet aspect.Les objectifs de ce travail, qui sont centrés sur les biais de non-réponse, étaient d’étudier l’apport de données supplémentaires (enquête complémentaire auprès de non-répondants et bases médico-administratives) et de discuter l’influence du taux de réponse sur l’erreur de non-réponse et l’erreur de mesure.L'analyse était centrée sur la surveillance épidémiologique des risques professionnels via l’exploitation des données de la phase pilote de la cohorte Coset-MSA à l’inclusion. Dans cette enquête, en plus des données recueillies par questionnaire (enquête initiale et enquête complémentaire auprès de non-répondants), des informations auxiliaires issues de bases médico-administratives (SNIIR-AM et MSA) étaient disponibles pour les répondants mais aussi pour les non-répondants à l’enquête par questionnaire.Les résultats montrent que les données de l’enquête initiale, qui présentait un taux de réponse de 24%, corrigées pour la non-réponse avec des informations auxiliaires directement liées à la thématique de l’enquête (la santé et le travail) fournissent des estimations de prévalence en général proches de celles obtenues grâce à la combinaison des données de l’enquête initiale et de l’enquête complémentaire (dont le taux de réponse atteignait 63%) après correction de la non réponse par ces mêmes informations auxiliaires. La recherche d'un taux de réponse maximal à l’aide d’une enquête complémentaire n’apparait donc pas nécessaire pour diminuer le biais de non réponse. Cette étude a néanmoins mis en avant l’existence de potentiels biais de mesure plus importants pour l’enquête initiale que pour l’enquête complémentaire. L’étude spécifique du compromis entre erreur de non-réponse et erreur de mesure montre que, pour les variables qui ont pu être étudiées, après correction de la non-réponse, la somme de l’erreur de non-réponse de l’erreur de mesure est équivalente dans l’enquête initiale et dans les enquêtes combinées (enquête initiale et complémentaire).Ce travail a montré l’intérêt des bases médico-administratives pour diminuer l’erreur de non-réponse et étudier les erreurs de mesure dans une enquête de surveillance épidémiologique
Nonresponse occurs in most epidemiologic surveys and may generate selection bias (which is, in this case, a nonresponse bias) when it is linked to outcome variables. In epidemiologic surveillance, whose one of the purpose is to estimate prevalences, it is usual to use survey sampling. In this case, unit nonresponse occurs and it is possible to use methods coming from survey sampling to correct for nonresponse. Nonresponse bias can be expressed as the product of the inverse of the response rate and the covariance between the probability of response and the outcome variable. Thus, two options are available to reduce the effect of nonresponse. The first is to increase the response rate by developing appropriate strategies at the study design phase. However, the maximization of the response rate can prompt other kinds of bias, such as measurement bias. In the second option, after data collection, information associated with both nonresponse and the outcome variable, and available for both respondents and nonrespondents, can be used to calculate corrective factors. This solution requires having information on the complete random sample (respondents and nonrespondents); but this information is rarely sufficient. Recent possibilities to access administrative databases (particularly those pertaining to health insurance) offer new perspectives on this aspect.The objectives of this work focused on the nonresponse bias were to study the contribution of supplementary data (administrative databases and complementary survey among nonrespondents) and to discuss the influence of the response rate on the nonresponse error and the measurement error. The analyses focused on occupational health epidemiologic surveillance, using data (at inclusion) from the Coset-MSA cohort pilot study. In this study, in addition to the data collected by questionnaire (initial and complementary survey among nonrespondents), auxiliary information from health and occupational administrative databases was available for both respondents and nonrespondents.Results show that the data from the initial survey (response rate : 24%), corrected for nonresponse with information directly linked to the study subject (health and work) produce estimations of prevalence close to those obtained by combining data from the initial survey and the complementary survey (response rate : 63%), after nonresponse adjustment on the same auxiliary information. Using a complementary survey to attain a maximal response rate does not seem to be necessary in order to decrease nonresponse bias. Nevertheless, this study highlights potential measurement bias which could be more consequential for the initial survey than for the complementary survey. The specific study of the trade-off between nonresponse error and measurement error shows that, for the studied variables and after correction for nonresponse, the sum of the nonresponse error and the measurement error is equivalent in the initial survey and in the combined surveys (initial plus complementary survey). This work illustrated the potential of administrative databases for decreasing the nonresponse error and for evaluating measurement error in an epidemiologic surveillance survey

By numerical methods, two models of anisotropic mixed quantum spin chains, consisting of spins of two different sizes, Sa = 1/2 and Sb = 1 as well as Sb = 3/2, are studied with respect to their critical properties at quantum phase transitions in a selected region of parameter space. The quantum spin chains are made up of basecells of four spins, according to the structure Sa − Sa − Sb − Sb. They are described by the XXZ Hamiltonian, that extends the quantum Heisenberg model by a variable anisotropic exchange interaction. As additional control parameter, an alternating exchange constant between nearest-neighbour spins is introduced. Insight gained by complementary application of exact diagonalization and quantum Monte Carlo simulations, as well as appropriate methods of analysis, is embedded in the broad existing knowledge on homogeneous quantum spin chains. In anisotropic homogeneous quantum spin chains, there exist phase boundaries with continuously varying critical exponents, the Gaussian critical lines, along which, in addition to standard scaling relations, further extended scaling relations hold. Reweighting methods, also applied to improved quantum Monte Carlo estimators, and finite-size scaling analysis of simulation data deliver a wealth of numerical results confirming the existence of a Gaussian critical line also in the mixed spin models considered. Extrapolation of exact data offers, apart from confirmation of simulation data, furthermore, insight into the conformal operator content of the model with Sb = 1
Mittels numerischer Methoden werden zwei Modelle anisotroper gemischter Quantenspinketten, bestehend aus Spins zweier unterschiedlicher Größen, Sa = 1/2 und Sb = 1 sowie Sb = 3/2, hinsichtlich ihrer kritischen Eigenschaften an Quanten-Phasenübergängen in einem ausgewählten Parameterbereich untersucht. Die Quantenspinketten sind aus Basiszellen zu vier Spins, gemäß der Struktur Sa − Sa − Sb − Sb, aufgebaut. Sie werden durch den XXZ Hamiltonoperator beschrieben, der das isotrope Quanten-Heisenberg Modell um eine variable anistrope Austauschwechselwirkung erweitert. Als zusätzlicher Kontrollparameter wird eine alterniernde Kopplungskonstante zwischen unmittelbar benachbarten Spins eingeführt. Die durch komplementäre Anwendung exakter Diagonalisierung und Quanten-Monte-Carlo Simulationen, sowie entsprechender Analyseverfahren, gewonnenen Erkenntnisse werden in das umfangreiche existierende Wissen über homogene Quantenspinketten eingebettet. Im Speziellen treten in anisotropen homogenen Quantenspinketten Phasengrenzen mit kontinuierlich variierenden kritischen Exponenten auf, die Gaußschen kritischen Linien, auf denen neben den herkömmlichen auch erweiterte Skalenrelationen Gültigkeit besitzen. Umgewichtungsmethoden, speziell auch angewandt auf verbesserte Quanten-Monte-Carlo Schätzer, und Endlichkeitsskalenanalyse von Simulationsdaten liefern eine Fülle von numerischen Ergebnissen, die das Auftreten der Gaußschen kritischen Linie auch in den untersuchten gemischten Quantenspinketten bestätigen. Die Extrapolation exakter Daten bietet, neben der Bestätigung der Simulationsdaten, darüber hinaus Einblick in einen Teil des konformen Operatorinhalts des Modells mit Sb = 1

Durante la última década y paralelamente al incremento de la velocidad de computación, las técnicas de simulación molecular se han erigido como una importante herramienta para la predicción de propiedades físicas de sistemas de interés industrial. Estas propiedades resultan esenciales en las industrias química y petroquímica a la hora de diseñar, optimizar, simular o controlar procesos. El actual coste moderado de computadoras potentes hace que la simulación molecular se convierta en una excelente opción para proporcionar predicciones de dichas propiedades. En particular, la capacidad predictiva de estas técnicas resulta muy importante cuando en los sistemas de interés toman parte compuestos tóxicos o condiciones extremas de temperatura o presión debido a la dificultad que entraña la experimentación a dichas condiciones. La simulación molecular proporciona una alternativa a los modelos termofísicos utilizados habitualmente en la industria como es el caso de las ecuaciones de estado, modelos de coeficientes de actividad o teorías de estados correspondientes, que resultan inadecuados al intentar reproducir propiedades complejas de fluidos como es el caso de las de fluidos que presentan enlaces de hidrógeno, polímeros, etc. En particular, los métodos de Monte Carlo (MC) constituyen, junto a la dinámica molecular, una de las técnicas de simulación molecular más adecuadas para el cálculo de propiedades termofísicas. Aunque, por contra del caso de la dinámica molecular, los métodos de Monte Carlo no proporcionan información acerca del proceso molecular o las trayectorias moleculares, éstos se centran en el estudio de propiedades de equilibrio y constituyen una herramienta, en general, más eficiente para el cálculo del equilibrio de fases o la consideración de sistemas que presenten elevados tiempos de relajación debido a su bajos coeficientes de difusión y altas viscosidades. Los objetivos de esta tesis se centran en el desarrollo y la mejora tanto de algoritmos de simulación como de potenciales intermoleculares, factor considerado clave para el desarrollo de las técnicas de simulación de Monte Carlo. En particular, en cuanto a los algoritmos de simulación, la localización de puntos críticos de una manera precisa ha constituido un problema para los métodos habitualmente utilizados en el cálculo de equlibrio de fases, como es el método del colectivo de GIBBS. La aparición de fuertes fluctuaciones de densidad en la región crítica hace imposible obtener datos de simulación en dicha región, debido al hecho de que las simulaciones son llevadas a cabo en una caja de simulación de longitud finita que es superada por la longitud de correlación. Con el fin de proporcionar una ruta adecuada para la localización de puntos críticos tanto de componentes puros como mezclas binarias, la primera parte de esta tesis está dedicada al desarrollo y aplicación de métodos adecuados que permitan superar las dificultades encontradas en el caso de los métodos convencionales. Con este fin se combinan estudios de escalado del tamaño de sitema con técnicas de "Histogram Reweighting" (HR). La aplicación de estos métodos se ha mostrado recientemente como mucho mejor fundamentada y precisa para el cálculo de puntos críticos de sistemas sencillos como es el caso del fluido de LennardJones (LJ). En esta tesis, estas técnicas han sido combinadas con el objetivo de extender su aplicación a mezclas reales de interés industrial. Previamente a su aplicación a dichas mezclas reales, el fluido de LennardJones, capaz de reproducir el comportamiento de fluidos sencillos como es el caso de argón o metano, ha sido tomado como referencia en un paso preliminar. A partir de simulaciones realizadas en el colectivo gran canónico y recombinadas mediante la mencionada técnica de "Histogram Reweighting" se han obtenido los diagramas de fases tanto de fluidos puros como de mezclas binarias. A su vez se han localizado con una gran precisión los puntos críticos de dichos sistemas mediante las técnicas de escalado del tamaño de sistema. Con el fin de extender la aplicación de dichas técnicas a sistemas multicomponente, se han introducido modificaciones a los métodos de HR evitando la construcción de histogramas y el consecuente uso de recursos de memoria. Además, se ha introducido una metodología alternativa, conocida como el cálculo del cumulante de cuarto orden o parámetro de Binder, con el fin de hacer más directa la localización del punto crítico. En particular, se proponen dos posibilidades, en primer lugar la intersección del parámetro de Binder para dos tamaños de sistema diferentes, o la intersección del parámetro de Binder con el valor conocido de la correspondiente clase de universalidad combinado con estudios de escalado. Por otro lado, y en un segundo frente, la segunda parte de esta tesis está dedicada al desarrollo de potenciales intermoleculares capaces de describir las energías inter e intramoleculares de las moléculas involucradas en las simulaciones. En la última década se han desarrolldo diferentes modelos de potenciales para una gran variedad de compuestos. Uno de los más comunmente utilizados para representar hidrocarburos y otras moléculas flexibles es el de átomos unidos, donde cada grupo químico es representado por un potencial del tipo de LennardJones. El uso de este tipo de potencial resulta en una significativa disminución del tiempo de cálculo cuando se compara con modelos que consideran la presencia explícita de la totalidad de los átomos. En particular, el trabajo realizado en esta tesis se centra en el desarrollo de potenciales de átomos unidos anisotrópicos (AUA), que se caracterizan por la inclusión de un desplazamiento de los centros de LennardJones en dirección a los hidrógenos de cada grupo, de manera que esta distancia se convierte en un tercer parámetro ajustable junto a los dos del potencial de LennardJones.
En la segunda parte de esta tesis se han desarrollado potenciales del tipo AUA4 para diferentes familias de compuesto que resultan de interés industrial como son los tiofenos, alcanoles y éteres. En el caso de los tiofenos este interés es debido a las cada vez más exigentes restricciones medioambientales que obligan a eliminar los compuestos con presencia de azufre. De aquí la creciente de necesidad de propiedades termodinámicas para esta familia de compuestos para la cual solo existe una cantidad de datos termodinámicos experimentales limitada. Con el fin de hacer posible la obtención de dichos datos a través de la simulación molecular hemos extendido el potencial intermolecular AUA4 a esta familia de compuestos. En segundo lugar, el uso de los compuestos oxigenados en el campo de los biocombustibles ha despertado un importante interés en la industria petroquímica por estos compuestos. En particular, los alcoholes más utilizados en la elaboración de los biocombustibles son el metanol y el etanol. Como en el caso de los tiofenos, hemos extendido el potencial AUA4 a esta familia de compuestos mediante la parametrización del grupo hidroxil y la inclusión de un grupo de cargas electrostáticas optimizadas de manera que reproduzcan de la mejor manera posible el potencial electrostático creado por una molecula de referencia en el vacío. Finalmente, y de manera análoga al caso de los alcanoles, el último capítulo de esta tesis la atención se centra en el desarrollo de un potencial AUA4 capaz de reproducir cuantitativamente las propiedades de coexistencia de la familia de los éteres, compuestos que son ampliamente utilizados como solventes.
Parallel with the increase of computer speed, in the last decade, molecular simulation techniques have emerged as important tools to predict physical properties of systems of industrial interest. These properties are essential in the chemical and petrochemical industries in order to perform process design, optimization, simulation and process control. The actual moderate cost of powerful computers converts molecular simulation into an excellent tool to provide predictions of such properties. In particular, the predictive capability of molecular simulation techniques becomes very important when dealing with extreme conditions of temperature and pressure as well as when toxic compounds are involved in the systems to be studied due to the fact that experimentation at such extreme conditions is difficult and expensive.
Consequently, alternative processes must be considered in order to obtain the required properties. Chemical and petrochemical industries have made intensive use of thermophysical models including equations of state, activity coefficients models and corresponding state theories. These predictions present the advantage of providing good approximations with minimal computational needs. However, these models are often inadequate when only a limited amount of information is available to determine the necesary parameters, or when trying to reproduce complex fluid properties such as that of molecules which exhibit hydrogen bonding, polymers, etc. In addition, there is no way for dynamical properties to be estimated in a consistent manner.
In this thesis, the HR and FSS techniques are combined with the main goal of extending the application of these methodologies to the calculation of the vaporliquid equilibrium and critical point of real mixtures. Before applying the methodologies to the real mixtures of industrial interest, the LennardJones fluid has been taken as a reference model and as a preliminary step. In this case, the predictions are affected only by the omnipresent statistical errors, but not by the accuracy of the model chosen to reproduce the behavior of the real molecules or the interatomic potential used to calculate the configurational energy of the system.
The simulations have been performed in the grand canonical ensemble (GCMC)using the GIBBS code. Liquidvapor coexistences curves have been obtained from HR techniques for pure fluids and binary mixtures, while critical parameters were obtained from FSS in order to close the phase envelope of the phase diagrams. In order to extend the calculations to multicomponent systems modifications to the conventional HR techniques have been introduced in order to avoid the construction of histograms and the consequent need for large memory resources. In addition an alternative methodology known as the fourth order cumulant calculation, also known as the Binder parameter, has been implemented to make the location of the critical point more straightforward. In particular, we propose the use of the fourth order cumulant calculation considering two different possibilities: either the intersection of the Binder parameter for two different system sizes or the intersection of the Binder parameter with the known value for the system universality class combined with a FSS study. The development of transferable potential models able to describe the inter and intramolecular energies of the molecules involved in the simulations constitutes an important field in the improvement of Monte Carlo techniques. In the last decade, potential models, also referred to as force fields, have been developed for a wide range of compounds. One of the most common approaches for modeling hydrocarbons and other flexible molecules is the use of the unitedatoms model, where each chemical group is represented by one LennardJones center. This scheme results in a significant reduction of the computational time as compared to allatoms models since the number of pair interactions goes as the square of the number of sites. Improvements on the standard unitedatoms model, where typically a 612 LennardJones center of force is placed on top of the most significant atom, have been proposed. For instance, the AUA model consists of a displacement of the LennardJones centers of force towards the hydrogen atoms, converting the distance of displacement into a third adjustable parameter. In this thesis we have developed AUA 4 intermolecular potentials for three different families of compounds. The family of ethers is of great importance due to their applications as solvents. The other two families, thiophenes and alkanols, play an important roles in the oil and gas industry. Thiophene due to current and future environmental restrictions and alkanols due ever higher importance and presence of biofuels in this industry.

碩士
國立臺灣科技大學
資訊工程系
93
In document retrieval systems, query terms play an important role which can affect the performance of document retrieval systems. The performance of document retrieval systems can be improved by using query terms expansion techniques and query terms reweighting techniques. In this thesis, we present two new methods for query terms expansion and query terms rewieghting. The first method chooses additional query terms for query expansion according to the degrees of importance of relevant terms and use fuzzy rules to infer their weights for document retrieval. The second method adjusts the weights of query terms to be optimal using neural networks for document retrieval. The proposed methods increase the performance of information retrieval systems for dealing with document retrieval.

碩士
國立中正大學
電機工程研究所
107
There have been many transfer learning approaches using deep learning or GAN in recent years. Going back to the original goal of transfer learning is reducing training cost, there are still some researches focusing on more traditional ways which are not deep or GAN approaches. One of traditional approaches published in recent years with great accuracy uses second-order statistics. Thus, for utilizing second-order statistics and the classifier of that approach, this thesis tries to combine them with other approaches. Intuitively, using multiple approaches to align data can get higher accuracy than those only consider one approach. Reweighting is a rarely used alignment method in transfer learning. Therefore, this thesis presents that combining second-order statistics, reweighting and new kind of classifier can get higher accuracy. This study uses one of the most popular data sets, Office-Caltech 10, to be the training data, and compares the accuracy with the original approaches’. The conclusions of experiments are consistent with the expectations. However, if experiments without considering effects between approaches might cause Negative transfer.

碩士
國立臺灣大學
資訊管理學研究所
107
With the enormous development of online music services, people nowadays have access to a massive amount of music covering a wide range of genres. Most of the music in the market are unknown to the public, representing the heavy tail in the long tail distribution. Therefore, recommender systems are an important tool for people to find new music that draws interest. However, the majority of methods proposed have focused on improving overall recommendation accuracy, which often leads to recommending hit songs since hit songs have more interaction data with users. This is not practical for music nerds because they are very likely to have heard of the hit songs, or even not interested in hit songs for they have different tastes. Another interesting thing is that as these niche songs are less likely to attract listeners, the ones they attract perceive the songs as higher quality than the mainstream songs. Thus, providing recommendations of the hidden gems in the long tail helps foster customer confidence. It is called the cold start problem in recommendation system when only a little or no historical data available for new items/users. Typically additional content is needed to solve the cold start problem. In this work, we design a content-based model using a convolutional recurrent neural network (CRNN) to extract features from the audio file and then apply to it a reweighting method targeting skewed data that we proposed. The main idea is to raise the importance of lesser-known songs when training the model, in other words, to give larger penalties on them. We experiment with the architecture for our content-based model as well as the design of the reweighting function. Besides, we apply the reweighting method on other several models, demonstrating the effectiveness not only on content-based models but also collaborative filtering-based and hybrid ones. Experiments show that models with our proposed skew reweighting method significantly outperform those without reweighting on lesser-known songs.

碩士
國立臺灣科技大學
資訊工程系
105
In recent years, data volume is getting larger along with the fast development of Internet technologies. Some datasets contain a huge number of labels, dimensions and data points. As a result, some of them cannot be loaded by typical classifiers, and some of them require very long and unacceptable time for execution. Extreme multi-label classification is designed for these challenges. Extreme multi-label classification differs from traditional multi-label classification in a number of ways including the need for lower execution time, training at an extreme scale with millions of data points, features and labels, etc. In order to enhance the practicality, in this paper, we focus on designing an extreme multi-label classification approach which can be performed on a single personal computer. We devise a two-phase framework for dealing with the above issues. In the reweighting phase, the prediction precision is improved by paying more attention on hard-to-classify instances and increasing the diversity of the model. In the pretesting phase, trees with lower quality will be removed from the prediction model for reducing the model size and increasing the prediction precision. Experiments on real world datasets will verify that the proposed method is able to generate better prediction results and the model size is successfully shrunk down.

In this study, an efficient coupling between Monte Carlo (MC) molecular simulation and Darcy-scale flow in porous media is presented. The cell-centered finite difference method with a non-uniform rectangular mesh were used to discretize the simulation domain and solve the governing equations. To speed up the MC simulations, we implemented a recently developed scheme that quickly generates MC Markov chains out of pre-computed ones, based on the reweighting and reconstruction algorithm. This method astonishingly reduces the required computational time by MC simulations from hours to seconds. In addition, the reweighting and reconstruction scheme, which was originally designed to work with the LJ potential model, is extended to work with a potential model that accounts for the molecular quadrupole moment of fluids with non-spherical molecules such as CO2. The potential model was used to simulate the thermodynamic equilibrium properties for single-phase and two-phase systems using the canonical ensemble and the Gibbs ensemble, respectively. Comparing the simulation results with the experimental data showed that the implemented model has an excellent fit outperforming the standard LJ model. To demonstrate the strength of the proposed coupling in terms of computational time efficiency and numerical accuracy in fluid properties, various numerical experiments covering different compressible single-phase flow scenarios were conducted. The novelty in the introduced scheme is in allowing an efficient coupling of the molecular scale and Darcy scale in reservoir simulators. This leads to an accurate description of the thermodynamic behavior of the simulated reservoir fluids; consequently enhancing the confidence in the flow predictions in porous media.

By numerical methods, two models of anisotropic mixed quantum spin chains, consisting of spins of two different sizes, Sa = 1/2 and Sb = 1 as well as Sb = 3/2, are studied with respect to their critical properties at quantum phase transitions in a selected region of parameter space. The quantum spin chains are made up of basecells of four spins, according to the structure Sa − Sa − Sb − Sb. They are described by the XXZ Hamiltonian, that extends the quantum Heisenberg model by a variable anisotropic exchange interaction. As additional control parameter, an alternating exchange constant between nearest-neighbour spins is introduced. Insight gained by complementary application of exact diagonalization and quantum Monte Carlo simulations, as well as appropriate methods of analysis, is embedded in the broad existing knowledge on homogeneous quantum spin chains. In anisotropic homogeneous quantum spin chains, there exist phase boundaries with continuously varying critical exponents, the Gaussian critical lines, along which, in addition to standard scaling relations, further extended scaling relations hold. Reweighting methods, also applied to improved quantum Monte Carlo estimators, and finite-size scaling analysis of simulation data deliver a wealth of numerical results confirming the existence of a Gaussian critical line also in the mixed spin models considered. Extrapolation of exact data offers, apart from confirmation of simulation data, furthermore, insight into the conformal operator content of the model with Sb = 1.
Mittels numerischer Methoden werden zwei Modelle anisotroper gemischter Quantenspinketten, bestehend aus Spins zweier unterschiedlicher Größen, Sa = 1/2 und Sb = 1 sowie Sb = 3/2, hinsichtlich ihrer kritischen Eigenschaften an Quanten-Phasenübergängen in einem ausgewählten Parameterbereich untersucht. Die Quantenspinketten sind aus Basiszellen zu vier Spins, gemäß der Struktur Sa − Sa − Sb − Sb, aufgebaut. Sie werden durch den XXZ Hamiltonoperator beschrieben, der das isotrope Quanten-Heisenberg Modell um eine variable anistrope Austauschwechselwirkung erweitert. Als zusätzlicher Kontrollparameter wird eine alterniernde Kopplungskonstante zwischen unmittelbar benachbarten Spins eingeführt. Die durch komplementäre Anwendung exakter Diagonalisierung und Quanten-Monte-Carlo Simulationen, sowie entsprechender Analyseverfahren, gewonnenen Erkenntnisse werden in das umfangreiche existierende Wissen über homogene Quantenspinketten eingebettet. Im Speziellen treten in anisotropen homogenen Quantenspinketten Phasengrenzen mit kontinuierlich variierenden kritischen Exponenten auf, die Gaußschen kritischen Linien, auf denen neben den herkömmlichen auch erweiterte Skalenrelationen Gültigkeit besitzen. Umgewichtungsmethoden, speziell auch angewandt auf verbesserte Quanten-Monte-Carlo Schätzer, und Endlichkeitsskalenanalyse von Simulationsdaten liefern eine Fülle von numerischen Ergebnissen, die das Auftreten der Gaußschen kritischen Linie auch in den untersuchten gemischten Quantenspinketten bestätigen. Die Extrapolation exakter Daten bietet, neben der Bestätigung der Simulationsdaten, darüber hinaus Einblick in einen Teil des konformen Operatorinhalts des Modells mit Sb = 1.