Search

Reinforcement learning is a powerful framework aiming to determine optimal behavior in highly complex decision-making scenarios. This objective can be achieved using policy iteration, which requires to solve a typically large linear system of equations. We propose the variational quantum policy iteration (VarQPI) algorithm, realizing this step with a NISQ-compatible quantum-enhanced subroutine. Its scalability is supported by an analysis of the structure of generic reinforcement learning environments, laying the foundation for potential quantum advantage with utility-scale quantum computers. Furthermore, we introduce the warm-start initialization variant (WS-VarQPI) that significantly reduces resource overhead. The algorithm solves a large FrozenLake environment with an underlying 256x256-dimensional linear system, indicating its practical robustness., Comment: This work has been submitted to the IEEE for possible publication. Copyright may be transferred without notice, after which this version may no longer be accessible. 9 pages, 6 figures, 1 table

Linear systems of equations can be found in various mathematical domains, as well as in the field of machine learning. By employing noisy intermediate-scale quantum devices, variational solvers promise to accelerate finding solutions for large systems. Although there is a wealth of theoretical research on these algorithms, only fragmentary implementations exist. To fill this gap, we have developed the variational-lse-solver framework, which realizes existing approaches in literature, and introduces several enhancements. The user-friendly interface is designed for researchers that work at the abstraction level of identifying and developing end-to-end applications., Comment: This work has been submitted to the IEEE for possible publication. Copyright may be transferred without notice, after which this version may no longer be accessible. 3 pages, 2 figures, 1 table

In der digitalen Forensik als auch in der akademischen Forschung zur natürlichen Sprache ist die Fragestellung der automatischen Feststellung der Urheberschaft von Dokumenten ein relevantes Thema. Dabei werden stilistische Merkmale von Autor:innen analysiert, anhand deren man den Ursprung zuordnen kann. Eine der bisher unbeantworteten Fragen in diesem Forschungsgebiet ist ob mehrsprachige Autor:innen für sie typische Merkmale in mehreren Sprachen verwenden, und ob man solche Merkmale für eine sprachübergreifende Analyse der Urheberschaft verwenden kann. Diese Arbeit beschäftigt sich mit diesem sprachenübergreifenden Szenario, und beinhaltet wissenschaftliche Beiträge in mehreren relevanten Bereichen. Ein großes Problem in diesem Forschungsfeld ist der Mangel an Datensätzen die für diese Art von Forschung verwendet werden können. In dieser Arbteit wird eine Methode präsentiert die durch die Verwendung von Kommentaren aus einer Social Media Plattform Datensätze in verschiedenen Sprachkombinationen zusammenstellt, die für die sprachenübergreifende Forschung von Urheberschaft verwendet werden kann. Desweiteren wird ein neues Merkmal vorgestellt welches für das automatische Machine Learning verwendet werden kann: DT-grams verwenden universelle grammatikale Eigenschaften von Texten, die sprachunabhängig berechnet werden können und vor allem in Szenarien mit wenig verfügbaren Daten effiziente Klassifizierungen ermöglichen. Durch die Entwicklung eines ausgiebigen Benchmarks für Urheberschaftsforschung wird ein weiterer Kontext für die Leistung von DT-grams und auch anderen etablierten Methoden geschaffen. Schließlich werden diese Ergebnisse durch Experimente in verwandten Disziplinen wie dem Authorship Profiling ergänzt., Determining the authorship of a document by analyzing the stylistic choices of authors can be used in digital forensics and is an important field in natural language processing. For authors that write in more than one language, one question that arises is which features of the written texts are being transferred to the respective other language, and if those features can be used to identify the author, independent from the language the features are learned from. This thesis focuses on this cross-language scenario, and presents contributions in multiple aspects. One big problem in this field is the availability of suitable datasets, since authors writing in multiple languages are relatively scarce. Leveraging internet-scale social media comments, a method is presented that composes datasets using multilingual authors, enabling true cross-language authorship research. In another contribution, a novel type of machine learning feature for cross-language analyses is presented: DT-grams are based on universal grammar features and can be used to effectively classif yauthors in small-scale attribution problems. Finally, to provide more context to the performance of the DT-grams in other fields, a benchmark for authorship attribution in general is presented, and also experiments in related fields such as authorship profiling are performed., Benjamin Murauer, Zusammenfassung in deutscher Sprache, Dissertation University of Innsbruck 2022

In der digitalen Forensik als auch in der akademischen Forschung zur natürlichen Sprache ist die Fragestellung der automatischen Feststellung der Urheberschaft von Dokumenten ein relevantes Thema. Dabei werden stilistische Merkmale von Autor:innen analysiert, anhand deren man den Ursprung zuordnen kann. Eine der bisher unbeantworteten Fragen in diesem Forschungsgebiet ist ob mehrsprachige Autor:innen für sie typische Merkmale in mehreren Sprachen verwenden, und ob man solche Merkmale für eine sprachübergreifende Analyse der Urheberschaft verwenden kann. Diese Arbeit beschäftigt sich mit diesem sprachenübergreifenden Szenario, und beinhaltet wissenschaftliche Beiträge in mehreren relevanten Bereichen. Ein großes Problem in diesem Forschungsfeld ist der Mangel an Datensätzen die für diese Art von Forschung verwendet werden können. In dieser Arbteit wird eine Methode präsentiert die durch die Verwendung von Kommentaren aus einer Social Media Plattform Datensätze in verschiedenen Sprachkombinationen zusammenstellt, die für die sprachenübergreifende Forschung von Urheberschaft verwendet werden kann. Desweiteren wird ein neues Merkmal vorgestellt welches für das automatische Machine Learning verwendet werden kann: DT-grams verwenden universelle grammatikale Eigenschaften von Texten, die sprachunabhängig berechnet werden können und vor allem in Szenarien mit wenig verfügbaren Daten effiziente Klassifizierungen ermöglichen. Durch die Entwicklung eines ausgiebigen Benchmarks für Urheberschaftsforschung wird ein weiterer Kontext für die Leistung von DT-grams und auch anderen etablierten Methoden geschaffen. Schließlich werden diese Ergebnisse durch Experimente in verwandten Disziplinen wie dem Authorship Profiling ergänzt., Determining the authorship of a document by analyzing the stylistic choices of authors can be used in digital forensics and is an important field in natural language processing. For authors that write in more than one language, one question that arises is which features of the written texts are being transferred to the respective other language, and if those features can be used to identify the author, independent from the language the features are learned from. This thesis focuses on this cross-language scenario, and presents contributions in multiple aspects. One big problem in this field is the availability of suitable datasets, since authors writing in multiple languages are relatively scarce. Leveraging internet-scale social media comments, a method is presented that composes datasets using multilingual authors, enabling true cross-language authorship research. In another contribution, a novel type of machine learning feature for cross-language analyses is presented: DT-grams are based on universal grammar features and can be used to effectively classif yauthors in small-scale attribution problems. Finally, to provide more context to the performance of the DT-grams in other fields, a benchmark for authorship attribution in general is presented, and also experiments in related fields such as authorship profiling are performed., Benjamin Murauer, Zusammenfassung in deutscher Sprache, Dissertation University of Innsbruck 2022

In the last decade, machine translation has become a popular means to deal with multilingual digital content. By providing higher quality translations, obfuscating the source language of a text becomes more attractive. In this paper, we analyze the ability to detect the source language from the translated output of two widely used commercial machine translation systems by utilizing machine-learning algorithms with basic textual features like n-grams. Evaluations show that the source language can be reconstructed with high accuracy for documents that contain a sufficient amount of translated text. In addition, we analyze how the document size influences the performance of the prediction, as well as how limiting the set of possible source languages improves the classification accuracy., Comment: This was peer-reviewed, accepted and presented at https://www.cicling.org/2018/, but the organizer somehow failed to publish the proceedings

Cross-language authorship attribution problems rely on either translation to enable the use of single-language features, or language-independent feature extraction methods. Until recently, the lack of datasets for this problem hindered the development of the latter, and single-language solutions were performed on machine-translated corpora. In this paper, we present a novel language-independent feature for authorship analysis based on dependency graphs and universal part of speech tags, called DT-grams (dependency tree grams), which are constructed by selecting specific sub-parts of the dependency graph of sentences. We evaluate DT-grams by performing cross-language authorship attribution on untranslated datasets of bilingual authors, showing that, on average, they achieve a macro-averaged F1 score of 0.081 higher than previous methods across five different language pairs. Additionally, by providing results for a diverse set of features for comparison, we provide a baseline on the previously undocumented task of untranslated cross-language authorship attribution., Comment: To be published in: "32. GI-Workshop Grundlagen von Datenbanken"

The contribution examines a long-standing lawsuit between canons and monks of the monastery of St. Ambrose. The different stages of the trial are clarified thanks to some unpublished documents, preserved in the Archives of the Basilica of St. Ambrose and in the State Archives of Milan. Even the appointment of delegated judges was not enough to resolve the case, which was finally discussed (and resolved) by Innocent III at the Apostolic See. An edition of the unpublished documents is in the appendix.

Jos. Murauer, Graz, Univ., Diss., 1888

Jos. Murauer, Graz, Univ., Diss., 1888

eingereicht von: Sebastiaan Murauer, Diplomarbeit Universität Innsbruck 2019

eingereicht von: Sebastiaan Murauer, Diplomarbeit Universität Innsbruck 2019