A detailed study has been made of the distribution of lead on a public shotgun range in the George Washington-Jefferson National Forests in southwestern Virginia. Sampling of more than 100 sites has yielded data on the distribution pattern of the lead shot. Since opening in 1993 through 2000, 11.1 metric tons of lead has accumulated over an area 220 by 300 m (66,000 m2) with an average rate of accumulation of 1.4 metric tons per year. More than 85 % of the total dispersed lead lies scattered in the forest that surrounds the approximately 60 by 60 m cleared shooting surface. Lead is irregularly distributed because of the use of stationary targets and the general trajectory of launched clay targets. Maximum concentrations occur at distances of about 28 m about 80 m, and at about 180 m reach a maximum value of more than 5000 g per m2. Significant amounts of fine particulate lead, generated during shooting and as a result of impact occur close to the shooting box but absent at distances beyond 50 m. / Master of Science

Estudamos transporte quântico dependente de spin em sistemas de ponto e de poço quântico acoplados a contatos magnéticos. O primeiro passo do nosso estudo foi a dedução de fórmulas originais para a corrente e para o ruído em sistemas com tunelamento dependente de spin, através do formalismo de funções de Green de mão equilíbrio. As equações deduzidas reproduzem casos limites da literatura - em particular as fórmulas de Landauer-Buttiker. Posteriormente aplicamos essas fórmulas para estudar três sistemas distintos: (1) ponto quântico acoplado a contatos ferromagnéticos, (2) um ponto quântico acoplado a múltiplos terminais ferromagnéticos, e (3) um poço quântico acoplado a terminais de semicondutor magnético diluído (DMS). No sistema (1) consideramos os alinhamentos paralelo (P) e anti-paralelo (AP) entre as magnetizações dos terminais. Nesse sistema levamos em conta interação de Coulomb e espalhamento de spin no ponto quântico. Com as fórmulas para corrente e ruído deduzidas aqui, encontramos, por exemplo, que a interação de Coulomb, combinada com o magnetismo dos eletrodos, leva a um bloqueio de Coulomb dependente de spin. Esse efeito por sua vez leva a uma polarização da corrente que pode ser modulada (intensidade e sinal) através de uma tens~ao externa. Também encontramos que o espalhamento de spin leva a comportamentos contrastantes entre corrente e ruído. Enquanto a corrente na configuração AP aumenta com a taxa de espalhamento de spin R, o ruído nessa mesma configuração é suprimido para uma certa faixa de valores de R. Um outro efeito interessante que observamos foi a possibilidade de se suprimir o ruído térmico através de uma tensão de porta. Para o sistema (2) mostramos que é possível injetar corrente &#8593-polarizada no ponto quântico e coletar simultaneamente correntes &#8593 e &#8595 polarizadas em terminais diferentes. Além disso, a corrente ao passar do reservatório emissor para um dos reservatórios coletores tem a sua polarização intensificada. Portanto esse sistema pode operar como inversor e amplificador de polarização de corrente. Por último, analisamos os efeitos de terminais DMS e quantização de Landau (na presença de um campo magnético externo) sobre a corrente e o ruído no sistema (3). Encontramos que o efeito Zeeman gigante nos terminais DMS, gerado pela interação de troca s-d, leva a uma polarização da corrente. Em particular, para uma certa faixa de tensão o efeito Zeeman gigante resulta na completa supressão de uma dada componente de spin no transporte. Com isso é possível controlar a polarização da corrente através de uma tensão externa. Também observamos oscilações na corrente, no ruído e no fator de Fano como função do campo magnético. / We study spin dependent quantum transport in quantum dots and quantum well devices attached to magnetic leads. We first derive general formulas, including electron-electron interaction and spin flip, for both current and noise, using the no equilibrium Green function technique (Keldysh). From our equations we regain limiting cases in the literature - in particular the Landauer-Buttiker formula when we neglect electron-electron interaction. We apply these formulas to study three distinct systems: (1) a quantum dot attached to two ferromagnetic leads, (2) a quantum dot linked to many ferromagnetic leads, and (3) a quantum well coupled to dilute magnetic semiconductor (DMS) terminals. In the first system we consider both parallel (P) and anti-parallel (AP) ferromagnetic alignments of the leads. Coulomb interaction and spin flip scattering are also taken into account. With the formulas for the current and the noise derived here, we find, for instance, that the Coulomb interaction, combined with the magnetism of the electrodes, gives rise to a spin-dependent Coulomb blockade. This effect allows the control (intensity and sign) of the current polarization via the bias voltage. We also observe that spin flip scattering yields contrasting behavior between current and shot noise. While the current in the AP configuration increases with the spin flip, the shot noise becomes suppressed for a range of spin flip rates. Another interesting finding is the possibility to suppress the thermal noise via a gate voltage. For the dot coupled to three magnetic leads, we show that it is possible to inject current &#8593-polarized into the dot from the FM emitter, detect simultaneously &#8593 and &#8595 - polarized currents at distinct collectors. In addition, we find that the current has its polarization amplified when going from the emitter to one of the collectors. Therefore we have a device that operates as both as current polarization inverter and amplifier. Finally, we analyze the effects of DMS leads and Landau quantization on the current and noise of system (3). We and that the giant Zeeman effect in the DMS leads, due to the it s-d exchange interaction, gives rise to a spin polarized current, and for a particular bias voltage range, full suppression of one spin component. This gives rise to the possibility of tuning the current polarization via the bias voltage. We also observe oscillations in the current, the noise and the Fano factor as a function of the magnetic field.

Il est important pour les robots de pouvoir reconnaître les objets rencontrés dans la vie quotidienne afin d’assurer leur autonomie. De nos jours, les robots sont équipés de capteurs sophistiqués permettant d’imiter le sens humain du toucher. C’est ce qui permet aux robots interagissant avec les objets de percevoir les propriétés (telles la texture, la rigidité et la matière) nécessaires pour leur reconnaissance. Dans cette thèse, notre but est d’exploiter les données haptiques issues de l’interaction robot-objet afin de reconnaître les objets de la vie quotidienne, et cela en utilisant les algorithmes d’apprentissage automatique. Le problème qui se pose est la difficulté de collecter suffisamment de données haptiques afin d’entraîner les algorithmes d’apprentissage supervisé sur tous les objets que le robot doit reconnaître. En effet, les objets de la vie quotidienne sont nombreux et l’interaction physique entre le robot et chaque objet pour la collection des données prend beaucoup de temps et d’efforts. Pour traiter ce problème, nous développons un système de reconnaissance haptique permettant de reconnaître des objets à partir d'aucune, de une seule, ou de plusieurs données d’entraînement. Enfin, nous intégrons la vision afin d’améliorer la reconnaissance d'objets lorsque le robot est équipé de caméras. / Recognizing surrounding objects is an important skill for the autonomy of robots performing in daily-life. Nowadays robots are equipped with sophisticated sensors imitating the human sense of touch. This allows the recognition of an object based on information ensuing from robot-object physical interaction. Such information can include the object texture, compliance and material. In this thesis, we exploit haptic data to perform haptic recognition of daily life objects using machine learning techniques. The main challenge faced in our work is the difficulty of collecting a fair amount of haptic training data for all daily-life objects. This is due to the continuously growing number of objects and to the effort and time needed by the robot to physically interact with each object for data collection. We solve this problem by developing a haptic recognition framework capable of performing Zero-shot, One-shot and Multi-shot Learning. We also extend our framework by integrating vision to enhance the robot’s recognition performance, whenever such sense is available.

Estudamos transporte quântico dependente de spin em sistemas de ponto e de poço quântico acoplados a contatos magnéticos. O primeiro passo do nosso estudo foi a dedução de fórmulas originais para a corrente e para o ruído em sistemas com tunelamento dependente de spin, através do formalismo de funções de Green de mão equilíbrio. As equações deduzidas reproduzem casos limites da literatura - em particular as fórmulas de Landauer-Buttiker. Posteriormente aplicamos essas fórmulas para estudar três sistemas distintos: (1) ponto quântico acoplado a contatos ferromagnéticos, (2) um ponto quântico acoplado a múltiplos terminais ferromagnéticos, e (3) um poço quântico acoplado a terminais de semicondutor magnético diluído (DMS). No sistema (1) consideramos os alinhamentos paralelo (P) e anti-paralelo (AP) entre as magnetizações dos terminais. Nesse sistema levamos em conta interação de Coulomb e espalhamento de spin no ponto quântico. Com as fórmulas para corrente e ruído deduzidas aqui, encontramos, por exemplo, que a interação de Coulomb, combinada com o magnetismo dos eletrodos, leva a um bloqueio de Coulomb dependente de spin. Esse efeito por sua vez leva a uma polarização da corrente que pode ser modulada (intensidade e sinal) através de uma tens~ao externa. Também encontramos que o espalhamento de spin leva a comportamentos contrastantes entre corrente e ruído. Enquanto a corrente na configuração AP aumenta com a taxa de espalhamento de spin R, o ruído nessa mesma configuração é suprimido para uma certa faixa de valores de R. Um outro efeito interessante que observamos foi a possibilidade de se suprimir o ruído térmico através de uma tensão de porta. Para o sistema (2) mostramos que é possível injetar corrente &#8593-polarizada no ponto quântico e coletar simultaneamente correntes &#8593 e &#8595 polarizadas em terminais diferentes. Além disso, a corrente ao passar do reservatório emissor para um dos reservatórios coletores tem a sua polarização intensificada. Portanto esse sistema pode operar como inversor e amplificador de polarização de corrente. Por último, analisamos os efeitos de terminais DMS e quantização de Landau (na presença de um campo magnético externo) sobre a corrente e o ruído no sistema (3). Encontramos que o efeito Zeeman gigante nos terminais DMS, gerado pela interação de troca s-d, leva a uma polarização da corrente. Em particular, para uma certa faixa de tensão o efeito Zeeman gigante resulta na completa supressão de uma dada componente de spin no transporte. Com isso é possível controlar a polarização da corrente através de uma tensão externa. Também observamos oscilações na corrente, no ruído e no fator de Fano como função do campo magnético. / We study spin dependent quantum transport in quantum dots and quantum well devices attached to magnetic leads. We first derive general formulas, including electron-electron interaction and spin flip, for both current and noise, using the no equilibrium Green function technique (Keldysh). From our equations we regain limiting cases in the literature - in particular the Landauer-Buttiker formula when we neglect electron-electron interaction. We apply these formulas to study three distinct systems: (1) a quantum dot attached to two ferromagnetic leads, (2) a quantum dot linked to many ferromagnetic leads, and (3) a quantum well coupled to dilute magnetic semiconductor (DMS) terminals. In the first system we consider both parallel (P) and anti-parallel (AP) ferromagnetic alignments of the leads. Coulomb interaction and spin flip scattering are also taken into account. With the formulas for the current and the noise derived here, we find, for instance, that the Coulomb interaction, combined with the magnetism of the electrodes, gives rise to a spin-dependent Coulomb blockade. This effect allows the control (intensity and sign) of the current polarization via the bias voltage. We also observe that spin flip scattering yields contrasting behavior between current and shot noise. While the current in the AP configuration increases with the spin flip, the shot noise becomes suppressed for a range of spin flip rates. Another interesting finding is the possibility to suppress the thermal noise via a gate voltage. For the dot coupled to three magnetic leads, we show that it is possible to inject current &#8593-polarized into the dot from the FM emitter, detect simultaneously &#8593 and &#8595 - polarized currents at distinct collectors. In addition, we find that the current has its polarization amplified when going from the emitter to one of the collectors. Therefore we have a device that operates as both as current polarization inverter and amplifier. Finally, we analyze the effects of DMS leads and Landau quantization on the current and noise of system (3). We and that the giant Zeeman effect in the DMS leads, due to the it s-d exchange interaction, gives rise to a spin polarized current, and for a particular bias voltage range, full suppression of one spin component. This gives rise to the possibility of tuning the current polarization via the bias voltage. We also observe oscillations in the current, the noise and the Fano factor as a function of the magnetic field.

One common shortcoming of modern computer vision is the inability of most models to generalize to new classes—one/few shot image recognition. We propose a new problem formulation for this task and present a network architecture and training methodology to solve this task. Further, we provide insights into how careful focus on how not just the data, but the way data presented to the model can have significant impact on performance. Using these method, we achieve high accuracy in few-shot image recognition tasks.

O aumento da vida em fadiga em peças submetidas a esforços cíclicos devido à aplicação de processos de tratamento mecânico superficial já é bastante conhecido, tanto no meio industrial quanto no meio acadêmico. Para molas, o processo de shot-peening se torna etapa essencial no processo de fabricação, porém um estudo sistemático do efeito do shot-peening na vida em fadiga se faz necessário. O objetivo deste trabalho é aprofundar o conhecimento nos processos de jateamento superficial do tipo shot-peening para molas em lâminas de veículos, através da análise de tensões residuais por difração de raios-x e ensaios de fadiga em uma série de amostras que sofreram dez diferentes receitas de processos de jateamento. Dos dez diferentes processos, o de jateamento com granalha esférica fundida de aço com 1,0mm de diâmetro seguido de um segundo jateamento com granalha esférica fundida de aço de 0,3mm de diâmetro levou a uma maior vida em fadiga das amostras. A análise por difração de raios-x comprovou que a importância das tensões residuais de compressão se dá até a uma profundidade de 0,05mm, influenciando diretamente no modo de nucleação de trincas de fadiga. Acima desta profundidade, as tensões residuais induzidas por shot-peening não têm influência no modo de propagação destas trincas, e por consequência na vida em fadiga das peças. Correlações entre os resultados e a alta dureza e diminuída ductilidade do material utilizado são discutidas. / The improvement of fatigue life in parts subjected to cyclic stresses by application of mechanical surface treatment processes is already well known, both in the industry and in the academy. Dealing with automotive springs, the shot peening process becomes an essential step in manufacturing these parts. In the case of leaf springs, however, a systematic investigation of the effect of shot peening on fatigue life is still required. The aim of this work is to improve the knowledge of shot peening on leaf springs for vehicles, through the analysis of residual stresses by x-ray diffraction and fatigue tests on a series of samples that were subjected to ten different peening schedules. Among the investigated processes, the usage of 1.0 mm diameter cast steel shot followed by a second peening with 0.3 mm diameter cast steel shot leads to better performance, regarding fatigue life. X-ray diffraction analysis shows that this improved performance may be attributed to residual compressive stress maintained until a depth of 0.05 mm below the surface, which directly influences the fatigue crack nucleation. Residual stresses induced by shot-peening in larger depths, have no influence on sample fatigue life, showing that crack propagation is not affected by the induced residual stresses. Consequently, the durability of parts is improved by shot-peening exclusively due to this influence on crack nucleation at samples surface. Correlations with the increased hardness and decreased ductility of the employed material are discussed.

O aumento da vida em fadiga em peças submetidas a esforços cíclicos devido à aplicação de processos de tratamento mecânico superficial já é bastante conhecido, tanto no meio industrial quanto no meio acadêmico. Para molas, o processo de shot-peening se torna etapa essencial no processo de fabricação, porém um estudo sistemático do efeito do shot-peening na vida em fadiga se faz necessário. O objetivo deste trabalho é aprofundar o conhecimento nos processos de jateamento superficial do tipo shot-peening para molas em lâminas de veículos, através da análise de tensões residuais por difração de raios-x e ensaios de fadiga em uma série de amostras que sofreram dez diferentes receitas de processos de jateamento. Dos dez diferentes processos, o de jateamento com granalha esférica fundida de aço com 1,0mm de diâmetro seguido de um segundo jateamento com granalha esférica fundida de aço de 0,3mm de diâmetro levou a uma maior vida em fadiga das amostras. A análise por difração de raios-x comprovou que a importância das tensões residuais de compressão se dá até a uma profundidade de 0,05mm, influenciando diretamente no modo de nucleação de trincas de fadiga. Acima desta profundidade, as tensões residuais induzidas por shot-peening não têm influência no modo de propagação destas trincas, e por consequência na vida em fadiga das peças. Correlações entre os resultados e a alta dureza e diminuída ductilidade do material utilizado são discutidas. / The improvement of fatigue life in parts subjected to cyclic stresses by application of mechanical surface treatment processes is already well known, both in the industry and in the academy. Dealing with automotive springs, the shot peening process becomes an essential step in manufacturing these parts. In the case of leaf springs, however, a systematic investigation of the effect of shot peening on fatigue life is still required. The aim of this work is to improve the knowledge of shot peening on leaf springs for vehicles, through the analysis of residual stresses by x-ray diffraction and fatigue tests on a series of samples that were subjected to ten different peening schedules. Among the investigated processes, the usage of 1.0 mm diameter cast steel shot followed by a second peening with 0.3 mm diameter cast steel shot leads to better performance, regarding fatigue life. X-ray diffraction analysis shows that this improved performance may be attributed to residual compressive stress maintained until a depth of 0.05 mm below the surface, which directly influences the fatigue crack nucleation. Residual stresses induced by shot-peening in larger depths, have no influence on sample fatigue life, showing that crack propagation is not affected by the induced residual stresses. Consequently, the durability of parts is improved by shot-peening exclusively due to this influence on crack nucleation at samples surface. Correlations with the increased hardness and decreased ductility of the employed material are discussed.

This dissertation explores the application of zero-shot text classification, a technique for categorizing texts without annotated data in the target domain. A true zero-shot setting breaks with the conventions of the traditional supervised machine learning paradigm that relies on quantitative in-domain evaluation for optimization, performance measurement, and model selection. The dissertation summarizes existing research to build a theoretical foundation for zero-shot methods, emphasizing efficiency and transparency. It benchmarks selected approaches across various tasks and datasets to understand their general performance, strengths, and weaknesses, mirroring the model selection process. On this foundation, two case studies demonstrate the application of zero-shot text classification: The first engages with historical German stock market reports, utilizing zero-shot methods for aspect-based sentiment classification. The case study reveals that although there are qualitative differences between finetuned and zero-shot approaches, the aggregated results are not easily distinguishable, sparking a discussion about the practical implications. The second case study integrates zero-shot text classification into a civil engineering document management system, showcasing how the flexibility of zero-shot models and the omission of the training process can benefit the development of prototype software, at the cost of an unknown performance. These findings indicate that, although zero-shot text classification works for the exemplary cases, the results are not generalizable. Taking up the findings of these case studies, the dissertation discusses dilemmas and theoretical considerations that arise from omitting the in-domain evaluation of applying zero-shot text classification. It concludes by advocating a broader focus beyond traditional quantitative metrics in order to build trust in zero-shot text classification, highlighting their practical utility as well as the necessity for further exploration as these technologies evolve.:1 Introduction 1.1 Problem Context 1.2 Related Work 1.3 Research Questions & Contribution 1.4 Author’s Publications 1.5 Structure of This Work 2 Research Context 2.1 The Current State of Text Classification 2.2 Efficiency 2.3 Approaches to Addressing Data Scarcity in Machine Learning 2.4 Challenges of Recent Developments 2.5 Model Sizes and Hardware Resources 2.6 Conclusion 3 Zero-shot Text Classification 3.1 Text Classification 3.2 State-of-the-Art in Text Classification 3.3 Neural Network Approaches to Data-Efficient Text Classification 3.4 Zero-shot Text Classification 3.5 Application 3.6 Requirements for Zero-shot Models 3.7 Approaches to Transfer Zero-shot 3.7.1 Terminology 3.7.2 Similarity-based and Siamese Networks 3.7.3 Language Model Token Predictions 3.7.4 Sentence Pair Classification 3.7.5 Instruction-following Models or Dialog-based Systems 3.8 Class Name Encoding in Text Classification 3.9 Approach Selection 3.10 Conclusion 4 Model Performance Survey 4.1 Experiments 4.1.1 Datasets 4.1.2 Model Selection 4.1.3 Hypothesis Templates 4.2 Zero-shot Model Evaluation 4.3 Dataset Complexity 4.4 Conclusion 5 Case Study: Historic German Stock Market Reports 5.1 Project 5.2 Motivation 5.3 Related Work 5.4 The Corpus and Dataset - Berliner Börsenzeitung 5.4.1 Corpus 5.4.2 Sentiment Aspects 5.4.3 Annotations 5.5 Methodology 5.5.1 Evaluation Approach 5.5.2 Trained Pipeline 5.5.3 Zero-shot Pipeline 5.5.4 Dictionary Pipeline 5.5.5 Tradeoffs 5.5.6 Label Space Definitions 5.6 Evaluation - Comparison of the Pipelines on BBZ 5.6.1 Sentence-based Sentiment 5.6.2 Aspect-based Sentiment 5.6.3 Qualitative Evaluation 5.7 Discussion and Conclusion 6 Case Study: Document Management in Civil Engineering 6.1 Project 6.2 Motivation 6.3 Related Work 6.4 The Corpus and Knowledge Graph 6.4.1 Data 6.4.2 BauGraph – The Knowledge Graph 6.5 Methodology 6.5.1 Document Insertion Pipeline 6.5.2 Frontend Integration 6.6 Discussion and Conclusion 7 MLMC 7.1 How it works 7.2 Motivation 7.3 Extensions of the Framework 7.4 Other Projects 7.4.1 Product Classification 7.4.2 Democracy Monitor 7.4.3 Climate Change Adaptation Finance 7.5 Conclusion 8 Discussion: The Five Dilemmas of Zero-shot 8.1 On Evaluation 8.2 The Five Dilemmas of Zero-shot 8.2.1 Dilemma of Evaluation or Are You Working at All? 8.2.2 Dilemma of Comparison or How Do I Get the Best Model? 8.2.3 Dilemma of Annotation and Label Definition or Are We Talking about the Same Thing? 8.2.4 Dilemma of Interpretation or Am I Biased? 8.2.5 Dilemma of Unsupervised Text Classification or Do I Have to Trust You? 8.3 Trust in Zero-shot Capabilities 8.4 Conclusion 9 Conclusion 9.1 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139 9.1.1 RQ1: Strengths and Weaknesses . . . . . . . . . . . . . . . . 139 9.1.2 RQ2: Application Studies . . . . . . . . . . . . . . . . . . . . 141 9.1.3 RQ3: Implications . . . . . . . . . . . . . . . . . . . . . . . . 143 9.2 Final Thoughts & Future Directions . . . . . . . . . . . . . . . . . . 144 References 147 A Appendix for Survey Chapter A.1 Model Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179 A.2 Task-specific Hypothesis Templates . . . . . . . . . . . . . . . . . . 180 A.3 Fractions of SotA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179 181 B Uncertainty vs. Accuracy 182 C Declaration of Authorship 185 D Declaration: Use of AI-Tools 186 E Bibliographic Data 187 / In dieser Dissertation wird die Anwendung von Zero-Shot-Textklassifikation -- die Kategorisierung von Texten ohne annotierte Daten in der Anwendungsdomäne -- untersucht. Ein echter Zero-Shot-Ansatz bricht mit den Konventionen des traditionellen überwachten maschinellen Lernens, welches auf einer quantitativen Evaluierung in der Zieldomäne zur Optimierung, Performanzmessung und Modellauswahl (model selection) basiert. Eine Zusammenfassung bestehender Forschungsarbeiten bildet die theoretische Grundlage für die verwendeten Zero-Shot-Methoden, wobei Effizienz und Transparenz im Vordergrund stehen. Ein Vergleich ausgewählter Ansätze mit verschiedenen Tasks und Datensätzen soll allgemeine Stärken und Schwächen aufzeigen und den Prozess der Modellauswahl widerspiegeln. Auf dieser Grundlage wird die Anwendung der Zero-Shot-Textklassifikation anhand von zwei Fallstudien demonstriert: Die erste befasst sich mit historischen deutschen Börsenberichten, wobei Zero-Shot zur aspekt-basierten Sentiment-Klassifikation eingesetzt wird. Es zeigt sich, dass es zwar qualitative Unterschiede zwischen trainierten und Zero-Shot-Ansätzen gibt, dass die aggregierten Ergebnisse aber nicht leicht zu unterscheiden sind, was Überlegungen zu praktischen Implikationen anstößt. Die zweite Fallstudie integriert Zero-Shot-Textklassifikation in ein Dokumentenmanagementsystem für das Bauwesen und zeigt, wie die Flexibilität von Zero-Shot-Modellen und der Wegfall des Trainingsprozesses die Entwicklung von Prototypen vereinfachen können -- mit dem Nachteil, dass die Genauigkeit des Modells unbekannt bleibt. Die Ergebnisse zeigen, dass die Zero-Shot-Textklassifikation in den Beispielanwendungen zwar annähernd funktioniert, die Ergebnisse aber nicht leicht verallgemeinerbar sind. Im Anschluss werden Dilemmata und theoretische Überlegungen erörtert, die sich aus dem Wegfall der Evaluierung in der Zieldomäne von Zero-Shot-Textklassifikation ergeben. Abschließend wird ein breiterer Fokus über die traditionellen quantitativen Metriken hinaus vorgeschlagen, um Vertrauen in die Zero-Shot-Textklassifikation aufzubauen und den praktischen Nutzen zu verbessern. Die Überlegungen zeigen aber auch die Notwendigkeit weiterer Forschung im Zuge der Weiterentwicklung dieser Technologien.:1 Introduction 1.1 Problem Context 1.2 Related Work 1.3 Research Questions & Contribution 1.4 Author’s Publications 1.5 Structure of This Work 2 Research Context 2.1 The Current State of Text Classification 2.2 Efficiency 2.3 Approaches to Addressing Data Scarcity in Machine Learning 2.4 Challenges of Recent Developments 2.5 Model Sizes and Hardware Resources 2.6 Conclusion 3 Zero-shot Text Classification 3.1 Text Classification 3.2 State-of-the-Art in Text Classification 3.3 Neural Network Approaches to Data-Efficient Text Classification 3.4 Zero-shot Text Classification 3.5 Application 3.6 Requirements for Zero-shot Models 3.7 Approaches to Transfer Zero-shot 3.7.1 Terminology 3.7.2 Similarity-based and Siamese Networks 3.7.3 Language Model Token Predictions 3.7.4 Sentence Pair Classification 3.7.5 Instruction-following Models or Dialog-based Systems 3.8 Class Name Encoding in Text Classification 3.9 Approach Selection 3.10 Conclusion 4 Model Performance Survey 4.1 Experiments 4.1.1 Datasets 4.1.2 Model Selection 4.1.3 Hypothesis Templates 4.2 Zero-shot Model Evaluation 4.3 Dataset Complexity 4.4 Conclusion 5 Case Study: Historic German Stock Market Reports 5.1 Project 5.2 Motivation 5.3 Related Work 5.4 The Corpus and Dataset - Berliner Börsenzeitung 5.4.1 Corpus 5.4.2 Sentiment Aspects 5.4.3 Annotations 5.5 Methodology 5.5.1 Evaluation Approach 5.5.2 Trained Pipeline 5.5.3 Zero-shot Pipeline 5.5.4 Dictionary Pipeline 5.5.5 Tradeoffs 5.5.6 Label Space Definitions 5.6 Evaluation - Comparison of the Pipelines on BBZ 5.6.1 Sentence-based Sentiment 5.6.2 Aspect-based Sentiment 5.6.3 Qualitative Evaluation 5.7 Discussion and Conclusion 6 Case Study: Document Management in Civil Engineering 6.1 Project 6.2 Motivation 6.3 Related Work 6.4 The Corpus and Knowledge Graph 6.4.1 Data 6.4.2 BauGraph – The Knowledge Graph 6.5 Methodology 6.5.1 Document Insertion Pipeline 6.5.2 Frontend Integration 6.6 Discussion and Conclusion 7 MLMC 7.1 How it works 7.2 Motivation 7.3 Extensions of the Framework 7.4 Other Projects 7.4.1 Product Classification 7.4.2 Democracy Monitor 7.4.3 Climate Change Adaptation Finance 7.5 Conclusion 8 Discussion: The Five Dilemmas of Zero-shot 8.1 On Evaluation 8.2 The Five Dilemmas of Zero-shot 8.2.1 Dilemma of Evaluation or Are You Working at All? 8.2.2 Dilemma of Comparison or How Do I Get the Best Model? 8.2.3 Dilemma of Annotation and Label Definition or Are We Talking about the Same Thing? 8.2.4 Dilemma of Interpretation or Am I Biased? 8.2.5 Dilemma of Unsupervised Text Classification or Do I Have to Trust You? 8.3 Trust in Zero-shot Capabilities 8.4 Conclusion 9 Conclusion 9.1 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139 9.1.1 RQ1: Strengths and Weaknesses . . . . . . . . . . . . . . . . 139 9.1.2 RQ2: Application Studies . . . . . . . . . . . . . . . . . . . . 141 9.1.3 RQ3: Implications . . . . . . . . . . . . . . . . . . . . . . . . 143 9.2 Final Thoughts & Future Directions . . . . . . . . . . . . . . . . . . 144 References 147 A Appendix for Survey Chapter A.1 Model Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179 A.2 Task-specific Hypothesis Templates . . . . . . . . . . . . . . . . . . 180 A.3 Fractions of SotA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179 181 B Uncertainty vs. Accuracy 182 C Declaration of Authorship 185 D Declaration: Use of AI-Tools 186 E Bibliographic Data 187

Audio description is considered a relatively new research field, and in South Africa it is not yet being used extensively to make visual channels accessible to blind audiences. Therefore research in this field is necessary. The literature review of this study shows that very little empirical research has been done where real audiences were tested. This study set out to test the impact of audio described film on the transportation, identification and comprehension of real audiences. The focus was on mise-en-shot elements specifically and the impact they have on the transportation, identification and comprehension of audiences. A thorough analysis of the two scenes and their audio descriptions showed that in the audio description the emphasis tends to be on the visual elements and not on the way these elements are presented (which forms part of mise-en-shot elements). The way the film is showed to the audience contributes to the film‟s meaning and the director often shows the film to the audience in a specific way to reach a certain effect or feeling in the film. If these mise-en-shot elements are not presented in the audio description of the film it, the blind audience cannot be expected to experience an equivalent effect to that of the sighted audience. In order to test the impact of mise-en-shot elements a methodology based on Bortolussi and Dixon‟s (2010) Psyconarratology was used. Immersion and transportation studies (Green and Brock, 2000 and Tal-Or & Cohen, 2010) were used to develop an experiment for the study. Two groups of respondents were tested by exposing one group to all the available channels of the film and another to only the audio description and soundtrack. The initial hypothesis of this study was that the audience exposed to the audio description and soundtrack of the film would experience less transportation, identification and comprehension due to the loss of mise-en-shot elements in the film. By using T-tests and qualitative comparisons of the responses it was found that the lack of mise-en-shot elements in the film had no statistically significant impact on the transportation of the audience. Furthermore it was found that the lack of mise-en-shot elements has an impact on the identification of the audience. The group that had access to all the channels were more able to identify with the characters. In terms of comprehension it was discovered that there were bigger differences between the two groups in the scenes were the director relied on the visual elements of the film. Thus there are opportunities for further research pertaining to the identification of the audience. Furthermore it is recommended that further research be done to include other film genres as well as alternative ways of doing audio description. South Africa as well as other countries can gain tremendously from the use of audio description to make visual channels accessible to blind audiences. To reach such a goal further research is needed in the field, not only to master the art of audio describing but to also create awareness for this aid. / Thesis (MA (Language Practice))--North-West University, Vaal Triangle Campus, 2013

Content-based video retrieval research combines expertise from many different areas, such as signal processing, machine learning, pattern recognition, and computer vision. As video extends into both the spatial and the temporal domain, we require techniques for the temporal decomposition of footage so that specific content can be accessed. This content may then be semantically classified - ideally in an automated process - to enable filtering, browsing, and searching. An important aspect that must be considered is that pictorial representation of information may be interpreted differently by individual users because it is less specific than its textual representation. In this thesis, we address several fundamental issues of content-based video retrieval for effective handling of digital footage. Temporal segmentation, the common first step in handling digital video, is the decomposition of video streams into smaller, semantically coherent entities. This is usually performed by detecting the transitions that separate single camera takes. While abrupt transitions - cuts - can be detected relatively well with existing techniques, effective detection of gradual transitions remains difficult. We present our approach to temporal video segmentation, proposing a novel algorithm that evaluates sets of frames using a relatively simple histogram feature. Our technique has been shown to range among the best existing shot segmentation algorithms in large-scale evaluations. The next step is semantic classification of each video segment to generate an index for content-based retrieval in video databases. Machine learning techniques can be applied effectively to classify video content. However, these techniques require manually classified examples for training before automatic classification of unseen content can be carried out. Manually classifying training examples is not trivial because of the implied ambiguity of visual content. We propose an unsupervised learning approach based on latent class modelling in which we obtain multiple judgements per video shot and model the users' response behaviour over a large collection of shots. This technique yields a more generic classification of the visual content. Moreover, it enables the quality assessment of the classification, and maximises the number of training examples by resolving disagreement. We apply this approach to data from a large-scale, collaborative annotation effort and present ways to improve the effectiveness for manual annotation of visual content by better design and specification of the process. Automatic speech recognition techniques along with semantic classification of video content can be used to implement video search using textual queries. This requires the application of text search techniques to video and the combination of different information sources. We explore several text-based query expansion techniques for speech-based video retrieval, and propose a fusion method to improve overall effectiveness. To combine both text and visual search approaches, we explore a fusion technique that combines spoken information and visual information using semantic keywords automatically assigned to the footage based on the visual content. The techniques that we propose help to facilitate effective content-based video retrieval and highlight the importance of considering different user interpretations of visual content. This allows better understanding of video content and a more holistic approach to multimedia retrieval in the future.