The rapid advancement of cutting-edge techniques has propelled state-of-the-art (SOTA) language models to new heights. Despite their impressive capabilities across a variety of downstream tasks, large language models still face many challenges such as hallucination and bias. The thesis focuses on two key objectives: first, it measures the robustness of T0_3B and investigates feasible methodologies to enhance the model’s robustness. Second, it targets on the explainability of large language models, aiming to make the intrinsic working mechanism more transparent and, consequently enhance model’s steerability. Motivated by the importance of mitigating non-robust behavior in language models, the thesis initially measures model’s robustness on handling minor perturbation. After that, I proposed and verified an approach to enhance robustness by making input more contextualized, a method that does not require the step of fine-tuning. Moreover, to understand the complex working mechanism of large language models, I designed and introduced two novel visualization tools: ’Logit Lens’ and ’Hidden States Plot in Spherical Coordinate System’. These tools, combined with additional experimental analysis, revealed a noticeable differentiation of the predicted processes between the first predicted token and subsequent tokens. The contributions of the thesis are mainly in the two following aspects: it provides feasible methodologies to enhance the robustness of language models without the need of fine-tuning, and it contributes to the field of explainable AI through the development of two visualization tools that shed light on the understanding of the working mechanism. / Den snabba utvecklingen av banbrytande tekniker har drivit språkmodeller till nya höjder. Trots deras imponerande prestanda över diverse språkrelaterade uppgifter, trots detta har dessa modeller fortfarande problem som hallucinationer och bias. Avhandlingen är centrerad kring två huvudmål: för det första undersöker den robustheten hos T0_3B och undersöker framtida strategier för att förbättra dess robusthet. För det andra utforskar den språkmodellernas ”förklaringsbarhet” (dvs hur väl vi förstår deras beteende), i syfte att göra dem mer transparenta och följaktligen förbättra modellens styrbarhet. Det första vi gör är att visa experiment som vi har satt upp för att mäta modellens robusthet mot mindre störningar. Som svar föreslår och underbygger vi ett tillvägagångssätt för att öka robustheten genom att ge modellen mer kontext när en fråga ställs, en metod som inte kräver vidare träning av modellen. Dessutom, för att förstå den komplexiteten hos språkmodeller, introducerar jag två nya visualiseringsverktyg: Logit Lens och Hidden States Plot i sfäriskt koordinatsystem. Dessa verktyg, i kombination med ytterligare experimentell analys, avslöjar ett diskting mönstr för den första förutspådda ordet jämfört med efterföljande ord. Bidragen från avhandlingen är huvudsakligen i de två följande aspekterna: den ger praktiska åtgärder för att förbättra robustheten hos språkmodeller utan behov av vidare träning, och den bidrar till området för förklarabar AI genom utvecklingen av två visualiseringsverktyg som ökar våran förståelse för hur dessa modeller fungerar.

Der in diesem Projekt erstellte digitale Zwilling eines 6-Achs-Roboters [1] soll zukünftig als Basis für Forschungsprojekte, insbesondere im Bereich maschinelles Lernen, der Erklärbarkeit von KI als auch des Einsatzes von innovativen Mensch-Technik-Interaktionen dienen. Dabei soll auch die Tauglichkeit von Unity als Plattform zur ganzheitlichen Umsetzung von digitalen Zwillingen ergründet werden. Dabei soll ein physikalisch repräsentatives Modell des realen Zwillings erstellt, und Kommunikation zwischen den Zwillingen realisiert werden. Motivation dafür ist, dass Unity alle nötigen Werkzeuge mitbringt um eine 3D-Visualisierung des Roboters, und der Umgebung zu erstellen, und über verknüpfte C#-Skripte auch Mittel zur Integrierung komplexer Programme liefert. Zu Simulationszwecken soll dem digitalen Zwilling eine selbstständige Steuerung verliehen werden. Dabei soll der Roboter anders als bei vielen existierenden Anwendung nicht nur durch simple Verschiebung der Bauteile anhand von simplen Bewegungsgleichungen, sondern durch Motorkräfte in seinen Gelenken angetrieben werden. Dazu wird die in Unity enthaltene Physik-Engine ohne Zusatzmodule verwendet. Außerdem soll der digitale Zwilling in der Lage sein durch Kommunikation mit dem realen Zwilling dessen Bewegungen zu verfolgen.

Systematic review of research manuscripts is a common procedure in which research studies pertaining a particular field or domain are classified and structured in a methodological way. This process involves, between other steps, an extensive review and consolidation of scientific metrics and attributes of the manuscripts, such as citations, type or venue of publication. The extraction and mapping of relevant publication data, evidently, is a very laborious task if performed manually. Automation of such systematic mapping steps intend to reduce the human effort required and therefore can potentially reduce the time required for this process.The objective of this thesis is to automate the data extraction and mapping steps when systematically reviewing studies. The manual process is replaced by novel graph modelling techniques for effective knowledge representation, as well as novel machine learning techniques that aim to learn these representations. This eventually automates this process by characterising the publications on the basis of certain sub-properties and qualities that give the reviewer a quick high-level overview of each research study. The final model is a concept learner that predicts these sub-properties which in addition addresses the inherent concept-drift of novel manuscripts over time. Different models were developed and explored in this research study for the development of concept learner.Results show that: (1) Graph reasoning techniques which leverage the expressive power in modern graph databases are very effective in capturing the extracted knowledge in a so-called knowledge graph, which allows us to form concepts that can be learned using standard machine learning techniques like logistic regression, decision trees and neural networks etc. (2) Neural network models and ensemble models outperformed other standard machine learning techniques like logistic regression and decision trees based on the evaluation metrics. (3) The concept learner is able to detect and avoid concept drift by retraining the model. / Systematisk granskning av forskningsmanuskript är en vanlig procedur där forskningsstudier inom ett visst område klassificeras och struktureras på ett metodologiskt sätt. Denna process innefattar en omfattande granskning och sammanförande av vetenskapliga mätvärden och attribut för manuskriptet, såsom citat, typ av manuskript eller publiceringsplats. Framställning och kartläggning av relevant publikationsdata är uppenbarligen en mycket mödosam uppgift om den utförs manuellt. Avsikten med automatiseringen av processen för denna typ av systematisk kartläggning är att minska den mänskliga ansträngningen, och den tid som krävs kan på så sätt minskas. Syftet med denna avhandling är att automatisera datautvinning och stegen för kartläggning vid systematisk granskning av studier. Den manuella processen ersätts av avancerade grafmodelleringstekniker för effektiv kunskapsrepresentation, liksom avancerade maskininlärningstekniker som syftar till att lära maskinen dessa representationer. Detta automatiserar så småningom denna process genom att karakterisera publikationerna beserat på vissa subjektiva egenskaper och kvaliter som ger granskaren en snabb god översikt över varje forskningsstudie. Den slutliga modellen är ett inlärningskoncept som förutsäger dessa subjektiva egenskaper och dessutom behandlar den inneboende konceptuella driften i manuskriptet över tiden. Olika modeller utvecklades och undersöktes i denna forskningsstudie för utvecklingen av inlärningskonceptet. Resultaten visar att: (1) Diagrammatiskt resonerande som uttnytjar moderna grafdatabaser är mycket effektiva för att fånga den framställda kunskapen i en så kallad kunskapsgraf, och gör det möjligt att vidareutveckla koncept som kan läras med hjälp av standard tekniker för maskininlärning. (2) Neurala nätverksmodeller och ensemblemodeller överträffade andra standard maskininlärningstekniker baserat på utvärderingsvärdena. (3) Inlärningskonceptet kan detektera och undvika konceptuell drift baserat på F1-poäng och omlärning av algoritmen.

The influence of Artificial Intelligence (AI) on society is increasing, with applications in highly sensitive and complicated areas. Examples include using Deep Convolutional Neural Networks within healthcare for diagnosing cancer. However, the inner workings of such models are often unknown, limiting the much-needed trust in the models. To combat this, Explainable AI (XAI) methods aim to provide explanations of the models' decision-making. Two such methods, Spectral Relevance Analysis (SpRAy) and Testing with Concept Activation Methods (TCAV), were evaluated on a deep learning model classifying cat and dog images that contained introduced artificial noise. The task was to assess the methods' capabilities to explain the importance of the introduced noise for the learnt model. The task was constructed as an exploratory step, with the future aim of using the methods on models diagnosing oral cancer. In addition to using the TCAV method as introduced by its authors, this study also utilizes the CAV-sensitivity to introduce and perform a sensitivity magnitude analysis. Both methods proved useful in discerning between the model’s two decision-making strategies based on either the animal or the noise. However, greater insight into the intricacies of said strategies is desired. Additionally, the methods provided a deeper understanding of the model’s learning, as the model did not seem to properly distinguish between the noise and the animal conceptually. The methods thus accentuated the limitations of the model, thereby increasing our trust in its abilities. In conclusion, the methods show promise regarding the task of detecting visually distinctive noise in images, which could extend to other distinctive features present in more complex problems. Consequently, more research should be conducted on applying these methods on more complex areas with specialized models and tasks, e.g. oral cancer.

Requirements analysis is an essential part of the System Development Life Cycle (SDLC) in order to achieve success in a software development project. There are several methods, techniques and frameworks used when expressing, prioritizing and managing requirements in IT projects. It is widely established that it is difficult to determine requirements for traditional systems, so a question naturally arises on how the requirements analysis is executed as AI solutions (that even fewer individuals can grasp) are being developed. Little research has been made on how the vital requirements phase is executed during development of AI solutions. This research aims to investigate the requirements analysis phase during the development of AI solutions. To explore this topic, an extensive literature review was made, and in order to collect new information, a number of interviews were performed with five suitable organizations (i.e, organizations that develop AI solutions). The results from the research concludes that the requirements analysis does not differ between development of AI solutions in comparison to development of traditional systems. However, the research showed that there were some deviations that can be deemed to be particularly unique for the development of AI solutions that affects the requirements analysis. These are: (1) the need for an iterative and agile systems development process, with an associated iterative and agile requirements analysis, (2) the importance of having a large set of quality data, (3) the relative deprioritization of user involvement, and (4) the difficulty of establishing timeframe, results/feasibility and the behavior of the AI solution beforehand.

As the field of Artificial Intelligence continues to evolve into a tool of societal impact, a need of breaking its initial boundaries as a computer science discipline arises to also include different humanistic fields. The work presented in this thesis revolves around the role that explainable artificial intelligence has in human-robot interaction through the study of poetry generators. To better understand the scope of the project, a poetry generators study presents the steps involved in the development process and the evaluation methods. In the algorithmic development of poetry generators, the shift from traditional disciplines to transdisciplinarity is identified. In collaboration with researchers from the Research Institutes of Sweden, state-of-the-art generators are tested to showcase the power of artificially enhanced artifacts. A development plateau is discovered and with the inclusion of Design Thinking methods potential future human-robot interaction development is identified. A physical prototype capable of verbal interaction on top of a poetry generator is created with the new feature of changing the corpora to any given audio input. Lastly, the strengths of transdisciplinarity are connected with the open-sourced community in regards to creativity and self-expression, producing an online tool to address future work improvements and introduce nonexperts to the steps required to self-build an intelligent robotic companion, thus also encouraging public technological literacy. Explainable AI is shown to help with user involvement in the process of creation, alteration and deployment of AI enhanced applications.

Articial intelligence has had a large impact on many industries and transformed some domains quite radically. There is tremendous potential in applying AI to the eld of medical diagnostics. A major issue with applying these techniques to some domains is an inability for AI models to provide an explanation or justication for their predictions. This creates a problem wherein a user may not trust an AI prediction, or there are legal requirements for justifying decisions that are not met. This thesis overviews how two explainable AI techniques (Shapley Additive Explanations and Local Interpretable Model-Agnostic Explanations) can establish a degree of trust for the user in the medical diagnostics eld. These techniques are evaluated through a user study. User study results suggest that supplementing classications or predictions with a post-hoc visualization increases interpretability by a small margin. Further investigation and research utilizing a user study surveyor interview is suggested to increase interpretability and explainability of machine learning results.

The deployment of human-augmented machine learning (ML) systems has become a recommended organizational best practice. ML systems use algorithms that rely on training data labeled by human annotators. However, human involvement in reviewing and labeling consumers' voice data to train speech recognition systems for Amazon Alexa, Microsoft Cortana, and the like has raised privacy concerns among consumers and privacy advocates. We use the enhanced APCO model as the theoretical lens to investigate how the disclosure of human involvement during the supervised machine learning process affects consumers' privacy decision making. In a scenario-based experiment with 499 participants, we present various company privacy policies to participants to examine their trust and privacy considerations, then ask them to share reasons why they would or would not opt in to share their voice data to train a companies' voice recognition software. We find that the perception of human involvement in the ML training process significantly influences participants' privacy-related concerns, which thereby mediate their decisions to share their voice data. Furthermore, we manipulate four factors of a privacy policy to operationalize various cognitive biases actively present in the minds of consumers and find that default trust and salience biases significantly affect participants' privacy decision making. Our results provide a deeper contextualized understanding of privacy-related concerns that may arise in human-augmented ML system configurations and highlight the managerial importance of considering the role of human involvement in supervised machine learning settings. Importantly, we introduce perceived human involvement as a new construct to the information privacy discourse. Although ubiquitous data collection and increased privacy breaches have elevated the reported concerns of consumers, consumers' behaviors do not always match their stated privacy concerns. Researchers refer to this as the privacy paradox, and decades of information privacy research have identified a myriad of explanations why this paradox occurs. Yet the underlying crux of the explanations presumes privacy concern to be the appropriate proxy to measure privacy attitude and compare with actual privacy behavior. Often, privacy concerns are situational and can be elicited through the setup of boundary conditions and the framing of different privacy scenarios. Drawing on the cognitive model of empowerment and interest, we propose a multidimensional privacy interest construct that captures consumers' situational and dispositional attitudes toward privacy, which can serve as a more robust measure in conditions leading to the privacy paradox. We define privacy interest as a consumer's general feeling toward reengaging particular behaviors that increase their information privacy. This construct comprises four dimensions—impact, awareness, meaningfulness, and competence—and is conceptualized as a consumer's assessment of contextual factors affecting their privacy perceptions and their global predisposition to respond to those factors. Importantly, interest was originally included in the privacy calculus but is largely absent in privacy studies and theoretical conceptualizations. Following MacKenzie et al. (2011), we developed and empirically validated a privacy interest scale. This study contributes to privacy research and practice by reconceptualizing a construct in the original privacy calculus theory and offering a renewed theoretical lens through which to view consumers' privacy attitudes and behaviors. / Doctor of Philosophy / The deployment of human-augmented machine learning (ML) systems has become a recommended organizational best practice. ML systems use algorithms that rely on training data labeled by human annotators. However, human involvement in reviewing and labeling consumers' voice data to train speech recognition systems for Amazon Alexa, Microsoft Cortana, and the like has raised privacy concerns among consumers and privacy advocates. We investigate how the disclosure of human involvement during the supervised machine learning process affects consumers' privacy decision making and find that the perception of human involvement in the ML training process significantly influences participants' privacy-related concerns. This thereby influences their decisions to share their voice data. Our results highlight the importance of understanding consumers' willingness to contribute their data to generate complete and diverse data sets to help companies reduce algorithmic biases and systematic unfairness in the decisions and outputs rendered by ML systems. Although ubiquitous data collection and increased privacy breaches have elevated the reported concerns of consumers, consumers' behaviors do not always match their stated privacy concerns. This is referred to as the privacy paradox, and decades of information privacy research have identified a myriad of explanations why this paradox occurs. Yet the underlying crux of the explanations presumes privacy concern to be the appropriate proxy to measure privacy attitude and compare with actual privacy behavior. We propose privacy interest as an alternative to privacy concern and assert that it can serve as a more robust measure in conditions leading to the privacy paradox. We define privacy interest as a consumer's general feeling toward reengaging particular behaviors that increase their information privacy. We found that privacy interest was more effective than privacy concern in predicting consumers' mobilization behaviors, such as publicly complaining about privacy issues to companies and third-party organizations, requesting to remove their information from company databases, and reducing their self-disclosure behaviors. By contrast, privacy concern was more effective than privacy interest in predicting consumers' behaviors to misrepresent their identity. By developing and empirically validating the privacy interest scale, we offer interest in privacy as a renewed theoretical lens through which to view consumers' privacy attitudes and behaviors.

Argumentation framework is a common technique in Artificial Intelligence and related fields. It is a good way of formalizing, resolving conflicts and helping with defeasible reasoning. This thesis discusses the exploration of the quantitative bipolar argumentation framework applied in multi-agent systems. Different agents in a multi-agent systems have various capabilities, and they contribute in different ways to the system goal. The purpose of this study is to explore approaches of explaining the overall behavior and output from a multi-agent system and enable explainability in the multi-agent systems. By exploring the properties of the quantitative bipolar argumentation framework using some techniques from explainable Artificial Intelligence (AI), the system will generate output with explanations given by the argumentation framework. This thesis gives a general overview of argumentation frameworks and common techniques from explainable AI. The study mainly focuses on the exploration of properties and interactive algorithms of quantitative bipolar argumentation framework. It introduces explanation techniques to the quantitative bipolar argumentation framework. A Graphical User Interface (GUI) application is included in order to present the results of the explanation. / Argumentationsramar är en vanlig teknik inom artificiell intelligens och relaterade områden. Det är ett bra sätt att formalisera, lösa konflikter och hjälpa till med defekta resonemang. I den här avhandlingen diskuteras utforskningen av den kvantitativa bipolära argumentationsramen som tillämpas i fleragentsystem. Olika agenter i ett system med flera agenter har olika kapacitet och bidrar på olika sätt till systemets mål. Syftet med den här studien är att utforska metoder för att förklara det övergripande beteendet och resultatet från ett system med flera agenter och möjliggöra förklarbarhet i systemen med flera agenter. Genom att utforska egenskaperna hos den kvantitativa bipolära argumentationsramen med hjälp av vissa tekniker från förklaringsbara AI kommer systemet att generera utdata med förklaringar som ges av argumentationsramen. Denna avhandling ger en allmän översikt över argumentationsramar och vanliga tekniker från förklaringsbara AI. Studien fokuserar främst på utforskandet av egenskaper och interaktiva algoritmer för det kvantitativa bipolära argumentationsramverket och introducerar tillämpningen av förklaringstekniker på det kvantitativa bipolära argumentationsramverket. En GUI-applikation ingår för att presentera resultaten av förklaringen.

The overprescription of antibiotics has resulted in bacteria resistance, which is considered a global threat to global health. Deciding if antibiotics should be prescribed or not from individual visits of patients’ medical records in Swedish can be considered a text classification task, one of the applications of Natural Language Processing (NLP). However, medical experts and patients can not trust a model if explanations for its decision are not provided. In this work, multilingual and monolingual Transformer models are evaluated for the medical classification task. Furthermore, local explanations are obtained with SHapley Additive exPlanations and Integrated Gradients to compare the models’ predictions and evaluate the explainability methods. Finally, the local explanations are also aggregated to obtain global explanations and understand the features that contributed the most to the prediction of each class. / Felaktig utskrivning av antibiotika har resulterat i ökad antibiotikaresistens, vilket anses vara ett globalt hot mot global hälsa. Att avgöra om antibiotika ska ordineras eller inte från patientjournaler på svenska kan betraktas som ett textklassificeringproblem, en av tillämpningarna av Natural Language Processing (NLP). Men medicinska experter och patienter kan inte lita på en modell om förklaringar till modellens beslut inte ges. I detta arbete utvärderades flerspråkiga och enspråkiga Transformersmodeller för medisinska textklassificeringproblemet. Dessutom erhölls lokala förklaringar med SHapley Additive exPlanations och Integrated gradients för att jämföra modellernas förutsägelser och utvärdera metodernas förklarbarhet. Slutligen aggregerades de lokala förklaringarna för att få globala förklaringar och förstå de ord som bidrog mest till modellens förutsägelse för varje klass.