Métodos de aprendizagem automática: um estudo baseado na avaliação e previsão de clientes bancários

Data Mining surge, hoje em dia, como uma ferramenta importante e crucial para o sucesso de um negócio. O considerável volume de dados que atualmente se encontra disponível, por si só, não traz valor acrescentado. No entanto, as ferramentas de Data Mining, capazes de transformar dados e mais dados em conhecimento, vêm colmatar esta lacuna, constituindo, assim, um trunfo que ninguém quer perder. O presente trabalho foca-se na utilização das técnicas de Data Mining no âmbito da atividade bancária, mais concretamente na sua atividade de telemarketing. Neste trabalho são aplicados catorze algoritmos a uma base de dados proveniente do call center de um banco português, resultante de uma campanha para a angariação de clientes para depósitos a prazo com taxas de juro favoráveis. Os catorze algoritmos aplicados no caso prático deste projeto podem ser agrupados em sete grupos: Árvores de Decisão, Redes Neuronais, Support Vector Machine, Voted Perceptron, métodos Ensemble, aprendizagem Bayesiana e Regressões. De forma a beneficiar, ainda mais, do que a área de Data Mining tem para oferecer, este trabalho incide ainda sobre o redimensionamento da base de dados em questão, através da aplicação de duas estratégias de seleção de atributos: Best First e Genetic Search. Um dos objetivos deste trabalho prende-se com a comparação dos resultados obtidos com os resultados presentes no estudo dos autores Sérgio Moro, Raul Laureano e Paulo Cortez (Sérgio Moro, Laureano, & Cortez, 2011). Adicionalmente, pretende-se identificar as variáveis mais relevantes aquando da identificação do potencial cliente deste produto financeiro. Como principais conclusões, depreende-se que os resultados obtidos são comparáveis com os resultados publicados pelos autores mencionados, sendo os mesmos de qualidade e consistentes. O algoritmo Bagging é o que apresenta melhores resultados e a variável referente à duração da chamada telefónica é a que mais influencia o sucesso de campanhas similares.

Mécanismes de déformation de nanoparticules d’Au par irradiation ionique

Résumé Dans la présente thèse, nous avons étudié la déformation anisotrope par bombardement ionique de nanoparticules d'or intégrées dans une matrice de silice amorphe ou d'arséniure d’aluminium cristallin. On s’est intéressé à la compréhension du mécanisme responsable de cette déformation pour lever toute ambigüité quant à l’explication de ce phénomène et pour avoir une interprétation consistante et unique. Un procédé hybride combinant la pulvérisation et le dépôt chimique en phase vapeur assisté par plasma a été utilisé pour la fabrication de couches nanocomposites Au/SiO2 sur des substrats de silice fondue. Des structures à couches simples et multiples ont été obtenues. Le chauffage pendant ou après le dépôt active l’agglomération des atomes d’Au et par conséquent favorise la croissance des nanoparticules. Les nanocomposites Au/AlAs ont été obtenus par implantation ionique de couches d’AlAs suivie de recuit thermique rapide. Les échantillons des deux nanocomposites refroidis avec de l’azote liquide ont été irradiés avec des faisceaux de Cu, de Si, d’Au ou d’In d’énergie allant de 2 à 40 MeV, aux fluences s'étendant de 1×1013 à 4×1015 ions/cm2, en utilisant le Tandem ou le Tandetron. Les propriétés structurales et morphologiques du nanocomposite Au/SiO2 sont extraites en utilisant des techniques optiques car la fréquence et la largeur de la résonance plasmon de surface dépendent de la forme et de la taille des nanoparticules, de leur concentration et de la distance qui les séparent ainsi que des propriétés diélectriques du matériau dans lequel les particules sont intégrées. La cristallinité de l’arséniure d’aluminium est étudiée par deux techniques: spectroscopie Raman et spectrométrie de rétrodiffusion Rutherford en mode canalisation (RBS/canalisation). La quantité d’Au dans les couches nanocomposites est déduite des résultats RBS. La distribution de taille et l’étude de la transformation de forme des nanoparticules métalliques dans les deux nanocomposites sont déterminées par microscopie électronique en transmission. Les résultats obtenus dans le cadre de ce travail ont fait l’objet de trois articles de revue. La première publication montre la possibilité de manipuler la position spectrale et la largeur de la bande d’absorption des nanoparticules d’or dans les nanocomposites Au/SiO2 en modifiant leur structure (forme, taille et distance entre particules). Les nanoparticules d’Au obtenues sont presque sphériques. La bande d’absorption plasmon de surface (PS) correspondante aux particules distantes est située à 520 nm. Lorsque la distance entre les particules est réduite, l’interaction dipolaire augmente ce qui élargit la bande de PS et la déplace vers le rouge (602 nm). Après irradiation ionique, les nanoparticules sphériques se transforment en ellipsoïdes alignés suivant la direction du faisceau. La bande d’absorption se divise en deux bandes : transversale et longitudinale. La bande correspondante au petit axe (transversale) est décalée vers le bleu et celle correspondante au grand axe (longitudinale) est décalée vers le rouge indiquant l’élongation des particules d’Au dans la direction du faisceau. Le deuxième article est consacré au rôle crucial de la déformation plastique de la matrice et à l’importance de la mobilité des atomes métalliques dans la déformation anisotrope des nanoparticules d’Au dans les nanocomposites Au/SiO2. Nos mesures montrent qu'une valeur seuil de 2 keV/nm (dans le pouvoir d'arrêt électronique) est nécessaire pour la déformation des nanoparticules d'or. Cette valeur est proche de celle requise pour la déformation de la silice. La mobilité des atomes d’Au lors du passage d’ions est confirmée par le calcul de la température dans les traces ioniques. Le troisième papier traite la tentative de formation et de déformation des nanoparticules d’Au dans une matrice d’arséniure d’aluminium cristallin connue pour sa haute résistance à l’amorphisation et à la déformation sous bombardement ionique. Le résultat principal de ce dernier article confirme le rôle essentiel de la matrice. Il s'avère que la déformation anisotrope du matériau environnant est indispensable pour la déformation des nanoparticules d’or. Les résultats expérimentaux mentionnés ci-haut et les calculs de températures dans les traces ioniques nous ont permis de proposer le scénario de déformation anisotrope des nanoparticules d’Au dans le nanocomposite Au/SiO2 suivant: - Chaque ion traversant la silice fait fondre brièvement un cylindre étroit autour de sa trajectoire formant ainsi une trace latente. Ceci a été confirmé par la valeur seuil du pouvoir d’arrêt électronique. - L’effet cumulatif des impacts de plusieurs ions conduit à la croissance anisotrope de la silice qui se contracte dans la direction du faisceau et s’allonge dans la direction perpendiculaire. Le modèle de chevauchement des traces ioniques (overlap en anglais) a été utilisé pour valider ce phénomène. - La déformation de la silice génère des contraintes qui agissent sur les nanoparticules dans les plans perpendiculaires à la trajectoire de l’ion. Afin d’accommoder ces contraintes les nanoparticules d’Au se déforment dans la direction du faisceau. - La déformation de l’or se produit lorsqu’il est traversé par un ion induisant la fusion d’un cylindre autour de sa trajectoire. La mobilité des atomes d’or a été confirmée par le calcul de la température équivalente à l’énergie déposée dans le matériau par les ions incidents. Le scénario ci-haut est compatible avec nos données expérimentales obtenues dans le cas du nanocomposite Au/SiO2. Il est appuyé par le fait que les nanoparticules d’Au ne se déforment pas lorsqu’elles sont intégrées dans l’AlAs résistant à la déformation.

Structures photoniques à base de nanocristaux de silicium

Il y a des indications que les nanocristaux de silicium (nc-Si) présentent un gain optique qui est potentiellement assez grand pour permettre l'amplification optique dans la gamme de longueurs d'ondes où une photoluminescence (PL) intense est mesurée (600- 1000 nm). Afin de fabriquer des cavités optiques, nous avons implantés des morceaux de silice fondue avec des ions de Si pour former une couche de nc-Si d'une épaisseur d'environ 1 μm. Le Si a été implanté à quatre énergies comprises entre 1 MeV et 1,9 MeV de manière à obtenir une concentration atomique de Si en excès variant entre 25% et 30%. Les pièces ont été flanquées de miroirs diélectriques composés de filtres interférentiels multicouches. Sur une plage de longueurs d'ondes d'environ 200 nm de large, un filtre réfléchit près de 100%, alors que l'autre a une réflexion moyenne d'environ 90%. Nous avons mesuré et comparé les spectres de PL de trois échantillons: le premier sans miroir, le second avec des filtres réfléchissant autour de 765 nm (entre 700 nm et 830 nm), et la troisième avec des filtres agissant autour de 875 nm (entre 810 nm et 940 nm). Lorsque les échantillons sont excités avec un laser pulsé à 390 nm, des mesures de photoluminescence résolue dans le temps (PLT) révèlent des taux de décroissance plus rapides en présence de miroirs dans le domaine de longueurs d'onde où ceux-ci agissent comparé aux échantillons sans miroirs. Aussi, l'intensité PL en fonction de la fluence d'excitation montre une augmentation plus rapide de la présence de miroirs, même si celle-ci reste sous-linéaire. Nous concluons que de l'émission stimulée pourrait être présente dans la cavité optique, mais sans dominer les autres mécanismes d'émission et de pertes.

La survenue des catastrophes naturelles : classification des variables explicatives par les réseaux de neurones

Durant les dernières décennies, l’occurrence des catastrophes naturelles a été fortement à la hausse. En effet, les catastrophes naturelles sont devenues de plus en plus fréquentes. En fait, ces risques dévastateurs ont touché durant les années précédentes différents pays dans des zones très diversifiées et continueront très probablement à être de réelles menaces dans le monde. Puisqu’aucun pays n’est à l’abri des catastrophes naturelles, il s’avère alors utile d’étudier les facteurs déterminants de leur survenue notamment avec la restriction de leurs périodes de retour et donc l’augmentation de leurs chances d’occurrence. Il nous a donc semblé opportun de tester les facteurs sous-jacents de la survenue des catastrophes naturelles. Notre travail se base sur l’application d’un réseau neuronal de type perceptron multicouche pour prédire le nombre des catastrophes naturelles à partir des variables les plus connues théoriquement. Ainsi, nous allons utiliser ce modèle neuronal pour effectuer l’analyse de sensitivité. Cette dernière permet de classer les variables explicatives selon l’importance de leur contribution dans la détermination du nombre de catastrophes naturelles comptabilisées durant la période d’étude. Les résultats obtenus ont montré que le réseau retenu peut prédire le nombre des catastrophes naturelles. De même, les différentes variables possèdent un effet considérable sur la sortie du réseau neuronal mais selon différents ordres d’importance. De ce fait, toutes ces variables contribuent à l’explication d’un problème aussi complexe comme la survenue des catastrophes naturelles.

Prévision du NO2 en utilisant la méthode du réseau de neurones

Cet article décrit une procédure de modélisation du phénomène de dispersion de la concentration du dioxyde de nitrogène (NO2) en utilisant la technique du perceptron multicouche (MLP). Notre objectif est de prouver que la concentration du NO2 est une variable autorégressive et expliquée par des variables météorologiques. Pour ce faire, nous avons procédé par trois étapes : dans la première étape nous avons utilisé la variable concentration NO2 uniquement, dans la seconde étape nous avons utilisé les variables météorologiques uniquement et dans la troisième étape nous avons utilisé la concentration du NO2 combinée avec les variables météorologiques. Les résultats ont montré que le troisième cas est plus performant que les deux autres ce qui prouve notre hypothèse.

A new method for load matching in multimode microwave feating applicators based on the use of dielectric layer superposition

Usage of a dielectric multilayer around a dielectric Sample is studied as a means for improving the efficiency in multimode microwave- heating cavities. The results show that by using additional dielectric constant layers the appearance of undesired reflections at the sample-air interface is avoided and higher power -absorption rates within the sample and high -efficiency designs are obtained

AN ARTIFICIAL NEURAL NETWORK FOR SONAR TARGET DETECTION

Neural Network has emerged as the topic of the day. The spectrum of its application is as wide as from ECG noise filtering to seismic data analysis and from elementary particle detection to electronic music composition. The focal point of the proposed work is an application of a massively parallel connectionist model network for detection of a sonar target. This task is segmented into: (i) generation of training patterns from sea noise that contains radiated noise of a target, for teaching the network;(ii) selection of suitable network topology and learning algorithm and (iii) training of the network and its subsequent testing where the network detects, in unknown patterns applied to it, the presence of the features it has already learned in. A three-layer perceptron using backpropagation learning is initially subjected to a recursive training with example patterns (derived from sea ambient noise with and without the radiated noise of a target). On every presentation, the error in the output of the network is propagated back and the weights and the bias associated with each neuron in the network are modified in proportion to this error measure. During this iterative process, the network converges and extracts the target features which get encoded into its generalized weights and biases.In every unknown pattern that the converged network subsequently confronts with, it searches for the features already learned and outputs an indication for their presence or absence. This capability for target detection is exhibited by the response of the network to various test patterns presented to it.Three network topologies are tried with two variants of backpropagation learning and a grading of the performance of each combination is subsequently made.

Characterization of Selected Photonic Materials and Systems Using Photoacoustic Technique

The photoacoustic investigations carried out on different photonic materials are presented in this thesis. Photonic materials selected for the investigation are tape cast ceramics, muItilayer dielectric coatings, organic dye doped PVA films and PMMA matrix doped with dye mixtures. The studies are performed by the measurement of photoacoustic signal generated as a result of modulated cw laser irradiation of samples. The gas-microphone scheme is employed for the detection of photoacoustic signal. The different measurements reported here reveal the adaptability and utility of the PA technique for the characterization of photonic materials.Ceramics find applications in the field of microelectronics industry. Tape cast ceramics are the building blocks of many electronic components and certain ceramic tapes are used as thermal barriers. The thermal parameters of these tapes will not be the same as that of thin films of the same materials. Parameters are influenced by the presence of foreign bodies in the matrix and the sample preparation technique. Measurements are done on ceramic tapes of Zirconia, Zirconia-Alumina combination, barium titanate, barium tin titanate, silicon carbide, lead zirconate titanateil'Z'T) and lead magnesium niobate titanate(PMNPT). Various configurations viz. heat reflection geometry and heat transmission geometry of the photoacoustic technique have been used for the evaluation of different thermal parameters of the sample. Heat reflection geometry of the PA cell has been used for the evaluation of thermal effusivity and heat transmission geometry has been made use of in the evaluation of thermal diffusivity. From the thermal diffusivity and thermal effusivity values, thermal conductivity is also calculated. The calculated values are nearly the same as the values reported for pure materials. This shows the feasibility of photoacoustic technique for the thermal characterization of ceramic tapes.Organic dyes find applications as holographic recording medium and as active media for laser operations. Knowledge of the photochemical stability of the material is essential if it has to be used tor any of these applications. Mixing one dye with another can change the properties of the resulting system. Through careful mixing of the dyes in appropriate proportions and incorporating them in polymer matrices, media of required stability can be prepared. Investigations are carried out on Rhodamine 6GRhodamine B mixture doped PMMA samples. Addition of RhB in small amounts is found to stabilize Rh6G against photodegradation and addition of Rh6G into RhB increases the photosensitivity of the latter. The PA technique has been successfully employed for the monitoring of dye mixture doped PMMA sample. The same technique has been used for the monitoring of photodegradation ofa laser dye, cresyl violet doped polyvinyl alcohol also.Another important application of photoacoustic technique is in nondestructive evaluation of layered samples. Depth profiling capability of PA technique has been used for the non-destructive testing of multilayer dielectric films, which are highly reflecting in the wavelength range selected for investigations. Eventhough calculation of thickness of the film is not possible, number of layers present in the system can be found out using PA technique. The phase plot has clear step like discontinuities, the number of which coincides with the number of layers present in the multilayer stack. This shows the sensitivity of PA signal phase to boundaries in a layered structure. This aspect of PA signal can be utilized in non-destructive depth profiling of reflecting samples and for the identification of defects in layered structures.

Surface studies of some metallic thin films by optical methods

The thesis aims to present the results of experimental investigations on the changes of optical properties of metallic thin films due to heating. The parameters which are measured are reflectivity, refractive indices and the ellipsometric quantities V and A . The materials used in the studies are metals like Silver, Aluminium and Copper. By applying the optical method the interdiffusion taking place in multilayer ‘films of Aluminium and Silver has also been studied. Special interest has been taken to reveal the mechanisms of the hillock growth and surface roughness caused by heating and their relation with the stress in the film

Tailoring Magnetic Properties of Co-Fe Thin Films by Topographical Modifications Using Thermal Annealing and Swift Heavy Ion Irradiation and Fabrication of Magnetoelectric Multilayers for Device Applications

Magnetism and magnetic materials have been playing a lead role in improving the quality of life. They are increasingly being used in a wide variety of applications ranging from compasses to modern technological devices. Metallic glasses occupy an important position among magnetic materials. They assume importance both from a scientific and an application point of view since they represent an amorphous form of condensed matter with significant deviation from thermodynamic equilibrium. Metallic glasses having good soft magnetic properties are widely used in tape recorder heads, cores of high-power transformers and metallic shields. Superconducting metallic glasses are being used to produce high magnetic fields and magnetic levitation effect. Upon heat treatment, they undergo structural relaxation leading to subtle rearrangements of constituent atoms. This leads to densification of amorphous phase and subsequent nanocrystallisation. The short-range structural relaxation phenomenon gives rise to significant variations in physical, mechanical and magnetic properties. Magnetic amorphous alloys of Co-Fe exhibit excellent soft magnetic properties which make them promising candidates for applications as transformer cores, sensors, and actuators. With the advent of microminiaturization and nanotechnology, thin film forms of these alloys are sought after for soft under layers for perpendicular recording media. The thin film forms of these alloys can also be used for fabrication of magnetic micro electro mechanical systems (magnetic MEMS). In bulk, they are drawn in the form of ribbons, often by melt spinning. The main constituents of these alloys are Co, Fe, Ni, Si, Mo and B. Mo acts as the grain growth inhibitor and Si and B facilitate the amorphous nature in the alloy structure. The ferromagnetic phases such as Co-Fe and Fe-Ni in the alloy composition determine the soft magnetic properties. The grain correlation length, a measure of the grain size, often determines the soft magnetic properties of these alloys. Amorphous alloys could be restructured in to their nanocrystalline counterparts by different techniques. The structure of nanocrystalline material consists of nanosized ferromagnetic crystallites embedded in an amorphous matrix. When the amorphous phase is ferromagnetic, they facilitate exchange coupling between nanocrystallites. This exchange coupling results in the vanishing of magnetocrystalline anisotropy which improves the soft magnetic properties. From a fundamental perspective, exchange correlation length and grain size are the deciding factors that determine the magnetic properties of these nanocrystalline materials. In thin films, surfaces and interfaces predominantly decides the bulk property and hence tailoring the surface roughness and morphology of the film could result in modified magnetic properties. Surface modifications can be achieved by thermal annealing at various temperatures. Ion irradiation is an alternative tool to modify the surface/structural properties. The surface evolution of a thin film under swift heavy ion (SHI) irradiation is an outcome of different competing mechanism. It could be sputtering induced by SHI followed by surface roughening process and the material transport induced smoothening process. The impingement of ions with different fluence on the alloy is bound to produce systematic microstructural changes and this could effectively be used for tailoring magnetic parameters namely coercivity, saturation magnetization, magnetic permeability and remanence of these materials. Swift heavy ion irradiation is a novel and an ingenious tool for surface modification which eventually will lead to changes in the bulk as well as surface magnetic property. SHI has been widely used as a method for the creation of latent tracks in thin films. The bombardment of SHI modifies the surfaces or interfaces or creates defects, which induces strain in the film. These changes will have profound influence on the magnetic anisotropy and the magnetisation of the specimen. Thus inducing structural and morphological changes by thermal annealing and swift heavy ion irradiation, which in turn induce changes in the magnetic properties of these alloys, is one of the motivation of this study. Multiferroic and magneto-electrics is a class of functional materials with wide application potential and are of great interest to material scientists and engineers. Magnetoelectric materials combine both magnetic as well as ferroelectric properties in a single specimen. The dielectric properties of such materials can be controlled by the application of an external magnetic field and the magnetic properties by an electric field. Composites with magnetic and piezo/ferroelectric individual phases are found to have strong magnetoelectric (ME) response at room temperature and hence are preferred to single phasic multiferroic materials. Currently research in this class of materials is towards optimization of the ME coupling by tailoring the piezoelectric and magnetostrictive properties of the two individual components of ME composites. The magnetoelectric coupling constant (MECC) (_ ME) is the parameter that decides the extent of interdependence of magnetic and electric response of the composite structure. Extensive investigates have been carried out in bulk composites possessing on giant ME coupling. These materials are fabricated by either gluing the individual components to each other or mixing the magnetic material to a piezoelectric matrix. The most extensively investigated material combinations are Lead Zirconate Titanate (PZT) or Lead Magnesium Niobate-Lead Titanate (PMNPT) as the piezoelectric, and Terfenol-D as the magnetostrictive phase and the coupling is measured in different configurations like transverse, longitudinal and inplane longitudinal. Fabrication of a lead free multiferroic composite with a strong ME response is the need of the hour from a device application point of view. The multilayer structure is expected to be far superior to bulk composites in terms of ME coupling since the piezoelectric (PE) layer can easily be poled electrically to enhance the piezoelectricity and hence the ME effect. The giant magnetostriction reported in the Co-Fe thin films makes it an ideal candidate for the ferromagnetic component and BaTiO3 which is a well known ferroelectric material with improved piezoelectric properties as the ferroelectric component. The multilayer structure of BaTiO3- CoFe- BaTiO3 is an ideal system to understand the underlying fundamental physics behind the ME coupling mechanism. Giant magnetoelectric coupling coefficient is anticipated for these multilayer structures of BaTiO3-CoFe-BaTiO3. This makes it an ideal candidate for cantilever applications in magnetic MEMS/NEMS devices. SrTiO3 is an incipient ferroelectric material which is paraelectric up to 0K in its pure unstressed form. Recently few studies showed that ferroelectricity can be induced by application of stress or by chemical / isotopic substitution. The search for room temperature magnetoelectric coupling in SrTiO3-CoFe-SrTiO3 multilayer structures is of fundamental interest. Yet another motivation of the present work is to fabricate multilayer structures consisting of CoFe/ BaTiO3 and CoFe/ SrTiO3 for possible giant ME coupling coefficient (MECC) values. These are lead free and hence promising candidates for MEMS applications. The elucidation of mechanism for the giant MECC also will be the part of the objective of this investigation.

Spontane Strukturbildung in Mehrschichtsystemen unter Verwendung Molekularer Gläser

Den Schwerpunkt dieser Dissertation bildet zum einen die Entwicklung eines theoretischen Modells zur Beschreibung des Strukturbildungsprozesses in organisch/anorganischen Doppelschichtsystemen und zum anderen die Untersuchung der Übertragbarkeit dieser theoretisch gewonnenen Ergebnisse auf reale Systeme. Hierzu dienen systematische experimentelle Untersuchungen dieses Phänomens an einem Testsystem. Der Bereich der selbstorganisierenden Systeme ist von hohem wissenschaftlichen Interesse, erlaubt er doch die Realisierung von Strukturen, die nicht den Begrenzungen heutiger Techniken unterliegen, wie etwa der Beugung bei lithographischen Verfahren. Darüber hinaus liefert ein vertieftes Verständnis des Strukturbildungsprozesses auch eine Möglichkeit, im Falle entsprechender technischer Anwendungen Instabilitäten innerhalb der Schichtsysteme zu verhindern und somit einer Degradation der Bauteile entgegenzuwirken. Im theoretischen Teil der Arbeit konnte ein Modell im Rahmen der klassischen Elastizitätstheorie entwickelt werden, mit dessen Hilfe sich die Entstehung der Strukturen in Doppelschichtsystemen verstehen läßt. Der hier gefundene funktionale Zusammenhang zwischen der Periode der Strukturen und dem Verhältnis der Schichtdicken von organischer und anorganischer Schicht, wird durch die experimentellen Ergebnisse sehr gut bestätigt. Die Ergebnisse zeigen, daß es technologisch möglich ist, über die Vorgabe der Schichtdicke in einem Materialsystem die Periodizität der entstehenden Strukturen vorzugeben. Darüber hinaus liefert das vorgestellte Modell eine Stabilitätsbedingung für die Schichtsysteme, die es ermöglicht, zu jedem Zeitpunkt die dominierende Mode zu identifizieren. Ein Schwerpunkt der experimentellen Untersuchungen dieser Arbeit liegt auf der Strukturbildung innerhalb der Schichtsysteme. Das Testsystem wurde durch Aufbringen einer organischen Schicht - eines sog. Molekularen Glases - auf ein Glassubstrat realisiert, als Deckschicht diente eine Siliziumnitrid-Schicht. Es wurden Proben mit variierenden Schichtdicken kontrolliert erwärmt. Sobald die Temperatur des Schichtsystems in der Größenordnung der Glasübergangstemperatur des jeweiligen organischen Materials lag, fand spontan eine Strukturbildung auf Grund einer Spannungsrelaxation statt. Es ließen sich durch die Wahl einer entsprechenden Heizquelle unterschiedliche Strukturen realisieren. Bei Verwendung eines gepulsten Lasers, also einer kreisförmigen Wärmequelle, ordneten sich die Strukturen konzentrisch an, wohingegen sich ihre Ausrichtung bei Verwendung einer flächenhaften Heizplatte statistisch verteilte. Auffällig bei allen Strukturen war eine starke Modulation der Oberfläche. Ferner konnte in der Arbeit gezeigt werden, daß sich durch eine gezielte Veränderung der Spannungsverteilung innerhalb der Schichtsysteme die Ausrichtung der Strukturen (gezielt) manipulieren ließen. Unabhängig davon erlaubte die Variation der Schichtdicken die Realisierung von Strukturen mit einer Periodizität im Bereich von einigen µm bis hinunter zu etwa 200 nm. Die Kontrolle über die Ausrichtung und die Periodizität ist Grundvoraussetzung für eine zukünftige technologische Nutzung des Effektes zur kontrollierten Herstellung von Mikro- bzw. Nanostrukturen. Darüber hinaus konnte ein zunächst von der Strukturbildung unabhängiges Konzept eines aktiven Sensors für die optische Raster-Nahfeld-Mikroskopie vorgestellt werden, das das oben beschriebene System, bestehend aus einem fluoreszierenden Molekularen Glas und einer Siliziumnitrid-Deckschicht, verwendet. Erste theoretische und experimentelle Ergebnisse zeigen das technologische Potential dieses Sensortyps.

Development of hybrid UV VCSEL with organic active material and dielectric DBR mirrors for medical, sensoric and data storage applications

Lasers play an important role for medical, sensoric and data storage devices. This thesis is focused on design, technology development, fabrication and characterization of hybrid ultraviolet Vertical-Cavity Surface-Emitting Lasers (UV VCSEL) with organic laser-active material and inorganic distributed Bragg reflectors (DBR). Multilayer structures with different layer thicknesses, refractive indices and absorption coefficients of the inorganic materials were studied using theoretical model calculations. During the simulations the structure parameters such as materials and thicknesses have been varied. This procedure was repeated several times during the design optimization process including also the feedback from technology and characterization. Two types of VCSEL devices were investigated. The first is an index coupled structure consisting of bottom and top DBR dielectric mirrors. In the space in between them is the cavity, which includes active region and defines the spectral gain profile. In this configuration the maximum electrical field is concentrated in the cavity and can destroy the chemical structure of the active material. The second type of laser is a so called complex coupled VCSEL. In this structure the active material is placed not only in the cavity but also in parts of the DBR structure. The simulations show that such a distribution of the active material reduces the required pumping power for reaching lasing threshold. High efficiency is achieved by substituting the dielectric material with high refractive index for the periods closer to the cavity. The inorganic materials for the DBR mirrors have been deposited by Plasma- Enhanced Chemical Vapor Deposition (PECVD) and Dual Ion Beam Sputtering (DIBS) machines. Extended optimizations of the technological processes have been performed. All the processes are carried out in a clean room Class 1 and Class 10000. The optical properties and the thicknesses of the layers are measured in-situ by spectroscopic ellipsometry and spectroscopic reflectometry. The surface roughness is analyzed by atomic force microscopy (AFM) and images of the devices are taken with scanning electron microscope (SEM). The silicon dioxide (SiO2) and silicon nitride (Si3N4) layers deposited by the PECVD machine show defects of the material structure and have higher absorption in the ultra violet range compared to ion beam deposition (IBD). This results in low reflectivity of the DBR mirrors and also reduces the optical properties of the VCSEL devices. However PECVD has the advantage that the stress in the layers can be tuned and compensated, in contrast to IBD at the moment. A sputtering machine Ionsys 1000 produced by Roth&Rau company, is used for the deposition of silicon dioxide (SiO2), silicon nitride (Si3N4), aluminum oxide (Al2O3) and zirconium dioxide (ZrO2). The chamber is equipped with main (sputter) and assisted ion sources. The dielectric materials were optimized by introducing additional oxygen and nitrogen into the chamber. DBR mirrors with different material combinations were deposited. The measured optical properties of the fabricated multilayer structures show an excellent agreement with the results of theoretical model calculations. The layers deposited by puttering show high compressive stress. As an active region a novel organic material with spiro-linked molecules is used. Two different materials have been evaporated by utilizing a dye evaporation machine in the clean room of the department Makromolekulare Chemie und Molekulare Materialien (mmCmm). The Spiro-Octopus-1 organic material has a maximum emission at the wavelength λemission = 395 nm and the Spiro-Pphenal has a maximum emission at the wavelength λemission = 418 nm. Both of them have high refractive index and can be combined with low refractive index materials like silicon dioxide (SiO2). The sputtering method shows excellent optical quality of the deposited materials and high reflection of the multilayer structures. The bottom DBR mirrors for all VCSEL devices were deposited by the DIBS machine, whereas the top DBR mirror deposited either by PECVD or by combination of PECVD and DIBS. The fabricated VCSEL structures were optically pumped by nitrogen laser at wavelength λpumping = 337 nm. The emission was measured by spectrometer. A radiation of the VCSEL structure at wavelength 392 nm and 420 nm is observed.

Spatio-temporal resolution studies on a highly compact ultrafast electron diffractometer

Time-resolved diffraction with femtosecond electron pulses has become a promising technique to directly provide insights into photo induced primary dynamics at the atomic level in molecules and solids. Ultrashort pulse duration as well as extensive spatial coherence are desired, however, space charge effects complicate the bunching of multiple electrons in a single pulse.Weexperimentally investigate the interplay between spatial and temporal aspects of resolution limits in ultrafast electron diffraction (UED) on our highly compact transmission electron diffractometer. To that end, the initial source size and charge density of electron bunches are systematically manipulated and the resulting bunch properties at the sample position are fully characterized in terms of lateral coherence, temporal width and diffracted intensity.Weobtain a so far not reported measured overall temporal resolution of 130 fs (full width at half maximum) corresponding to 60 fs (root mean square) and transversal coherence lengths up to 20 nm. Instrumental impacts on the effective signal yield in diffraction and electron pulse brightness are discussed as well. The performance of our compactUEDsetup at selected electron pulse conditions is finally demonstrated in a time-resolved study of lattice heating in multilayer graphene after optical excitation.

Building Grounded Abstractions for Artificial Intelligence Programming

Most Artificial Intelligence (AI) work can be characterized as either ``high-level'' (e.g., logical, symbolic) or ``low-level'' (e.g., connectionist networks, behavior-based robotics). Each approach suffers from particular drawbacks. High-level AI uses abstractions that often have no relation to the way real, biological brains work. Low-level AI, on the other hand, tends to lack the powerful abstractions that are needed to express complex structures and relationships. I have tried to combine the best features of both approaches, by building a set of programming abstractions defined in terms of simple, biologically plausible components. At the ``ground level'', I define a primitive, perceptron-like computational unit. I then show how more abstract computational units may be implemented in terms of the primitive units, and show the utility of the abstract units in sample networks. The new units make it possible to build networks using concepts such as long-term memories, short-term memories, and frames. As a demonstration of these abstractions, I have implemented a simulator for ``creatures'' controlled by a network of abstract units. The creatures exist in a simple 2D world, and exhibit behaviors such as catching mobile prey and sorting colored blocks into matching boxes. This program demonstrates that it is possible to build systems that can interact effectively with a dynamic physical environment, yet use symbolic representations to control aspects of their behavior.

Forecasting Global Temperature Variations by Neural Networks

Global temperature variations between 1861 and 1984 are forecast usingsregularization networks, multilayer perceptrons and linearsautoregression. The regularization network, optimized by stochasticsgradient descent associated with colored noise, gives the bestsforecasts. For all the models, prediction errors noticeably increasesafter 1965. These results are consistent with the hypothesis that thesclimate dynamics is characterized by low-dimensional chaos and thatsthe it may have changed at some point after 1965, which is alsosconsistent with the recent idea of climate change.s