Logo des Repositoriums

Auflistung nach Schlagwort "Neural Networks"

1 - 10 von 11
  • Zeitschriftenartikel
    Automatisierte nichtinvasive Emotionsmessung. Ein Erfahrungsbericht von der Vision bis zur Realisierung
    (i-com: Vol. 11, No. 3, 2012) Gerber, Paul; Ullrich, Daniel
    Affektive Reaktionen spielen eine wichtige Rolle in der Usability- und User Experience Forschung und sind häufig Gegens-tand in entsprechenden Messinstrumenten. Diese Messinstrumente haben aber oft den Nachteil, dass sie den Interaktions-prozess unterbrechen, etwa wenn sie während eines Usability-Tests durchgeführt werden. Oder aber sie unterliegen retro-spektiven Verfälschungen, zum Beispiel wenn sie nach Abschluss der Interaktion durchgeführt werden. Ziel unserer Arbeit war die Entwicklung eines nichtinvasiven Messinstruments, das emotionale Reaktionen während einer Interaktion erfassbar macht, ohne dass der Interaktionsprozess unterbrochen werden müsste. In dem vorliegenden Beitrag skizzieren wir den Prozess der Entwicklung dieses Messinstruments, von der ursprünglichen Vision bis zur Realisierung. Dabei gehen wir auf theoretische wie praktische Hindernisse ein und beschreiben, wie diese die finale Lösung beeinflusst haben. Abschließend berichten wir Ergebnisse zur Genauigkeit aus der P...
  • Konferenzbeitrag
    Development of neural network based rules for confusion set disambiguation in LanguageTool
    (SKILL 2018 - Studierendenkonferenz Informatik, 2018) Brenneis, Markus
    Confusion set disambiguation is a typical task for grammar checkers like LanguageTool. In this paper we present a neural network based approach which has low memory requirements, high precision with decent recall, and can easily be integrated into LanguageTool. Furthermore, adding support for new confusion pairs does not need any knowledge of the target language. We examine different sampling techniques and neural network architectures and compare our approaches with an existing memory-based algorithm.
  • Textdokument
    From Physical to Virtual: Leveraging Drone Imagery to Automate Photovoltaic System Maintenance
    (INFORMATIK 2021, 2021) Lowin, Maximilian; Kellner, Domenic; Kohl, Tobias; Mihale-Wilson, Cristina
    Optimizing the maintenance of large-scale infrastructure can be a significant cost driver for small and medium-sized enterprises (SMEs). This paper presents a feasible approach to combine data from real-world physical structures collected through an automated maintenance process with cloud-based AI services to generate a meaningful virtual representation of such structures. We use photovoltaic systems as an exemplary physical structure and thermal imaging, collected through scheduled drone monitoring. With help of these unstructured data sources, we demonstrate our approach's applicability. Our solution artifact provides a lightweight AI application that is adoptable for other problem spaces, enabling an easier knowledge transfer from research to SMEs. By combining Cloud Computing with Machine Learning, the artifact identifies present and emerging damages of physical objects. It provides a virtual representation of the object's state and empowers a meaningful visualization.
  • Zeitschriftenartikel
    Human Capacities for Emotion Recognition and their Implications for Computer Vision
    (i-com: Vol. 14, No. 2, 2015) Liebold, Benny; Richter, René; Teichmann, Michael; Hamker, Fred H.; Ohler, Peter
    Current models for automated emotion recognition are developed under the assumption that emotion expressions are distinct expression patterns for basic emotions. Thereby, these approaches fail to account for the emotional processes underlying emotion expressions. We review the literature on human emotion processing and suggest an alternative approach to affective computing. We postulate that the generalizability and robustness of these models can be greatly increased by three major steps: (1) modeling emotional processes as a necessary foundation of emotion recognition; (2) basing models of emotional processes on our knowledge about the human brain; (3) conceptualizing emotions based on appraisal processes and thus regarding emotion expressions as expressive behavior linked to these appraisals rather than fixed neuro-motor patterns. Since modeling emotional processes after neurobiological processes can be considered a long-term effort, we suggest that researchers should focus on early appraisals, which evaluate intrinsic stimulus properties with little higher cortical involvement. With this goal in mind, we focus on the amygdala and its neural connectivity pattern as a promising structure for early emotional processing. We derive a model for the amygdala-visual cortex circuit from the current state of neuroscientific research. This model is capable of conditioning visual stimuli with body reactions to enable rapid emotional processing of stimuli consistent with early stages of psychological appraisal theories. Additionally, amygdala activity can feed back to visual areas to modulate attention allocation according to the emotional relevance of a stimulus. The implications of the model considering other approaches to automated emotion recognition are discussed.
  • Textdokument
    Identifying a Trial Population for Clinical Studies on Diabetes Drug Testing with Neural Networks
    (SKILL 2021, 2021) Löhr, Tim
    This project aims to model an end-to-end workflow of implementing different Artificial Intelligence (AI) tools for a clinical environment. A possible use case, such as the selection process of patients for a novel treatment, will be conducted by estimating the hospitalization time with a Neural Network on an Electronic Health Record (EHR) of diabetes. Then, Explainable AI (XAI) methods are computed for models trained with a Random Forest to evaluate the predictions. The diabetes readmission EHR dataset from the University of California, Irvine (UCI) Diabetes is used for this project. The trial population is selected by predicting the expected days for a person being hospitalized. An arbitrary boundary is set for choosing whether or not a patient shall be included into the trial. If so, a clear explanation of how the prediction is calculated and additional possible risk factors will be given in order to make the workflow explainable. This project shows that given a proper explanatory approach, AI can be a useful tool for the modern clinical environment. The workflow finally reveals that AI can be a beneficial support tool for doctors in the patient selection process.
  • Workshopbeitrag
    Immersions - Musik in Künstlichen Ohren
    (Mensch und Computer 2019 - Workshopband, 2019) Herrmann, Vincent
    Immersions ist eine Performance Elektronischer Musik, die in Echtzeit die inneren Vorgänge eines Neuronalen Netzes sonifiziert. Dabei werden durch ein Optimierungsverfahren Klänge generiert, die bestimmte Bereiche des Netzwerks aktivieren. So wird hörbar, wie Musik in diesem künstlichen Ohr klingt. Zusätzlich werden die Vorgänge auch visualisiert, um eine möglichst immersive Erfahrung zu schaffen und es erlaubt Einzutauchen in die Tiefe Künstlicher Intelligenz.
  • Textdokument
    The Impact of Domain Knowledge on Applying Machine Learning Methods to Exoplanet Detection
    (SKILL 2021, 2021) Nguyen, The-Gia Leo
    Exoplanets do not emit electromagnetic waves which makes it challenging to detect them. Based on transit photometry, we trained a neural network on NASA Kepler space telescope data to detect exoplanets based on light intensity curves. We showcase, that with a well designed data pipeline, a small neural network is sufficient to achieve state-of-the-art performance, saving both computation time and hardware cost. The strongest improvement in performance could only be achieved by adding domain specific processing steps to the data pipeline. Domain knowledge was essential in selecting the appropriate machine learning concepts that are beneficial to solving the problem and have a higher impact on the performance than the actual classification method itself. We encourage to consider the data pipeline as an additional component, besides the classification model, that can potentially improve the overall performance.
  • Konferenzbeitrag
    NN2SQL: Let SQL Think for Neural Networks
    (BTW 2023, 2023) Schüle, Maximilian Emanuel; Kemper, Alfons; Neumann, Thomas
    Although database systems perform well in data access and manipulation, their relational model hinders data scientists from formulating machine learning algorithms in SQL. Nevertheless, we argue that modern database systems perform well for machine learning algorithms expressed in relational algebra. To overcome the barrier of the relational model, this paper shows how to transform data into a relational representation for training neural networks in SQL: We first describe building blocks for data transformation in SQL. Then, we compare an implementation for model training using array data types to the one using a relational representation in SQL-92 only. The evaluation proves the suitability of modern database systems for matrix algebra, although specialised array data types perform better than matrices in relational representation.
  • Textdokument
    Paving the road to a circular textile economy with AI
    (INFORMATIK 2021, 2021) Rudisch, Katharina; Jüngling, Sebastian; Carrillo Mendoza, Ricardo; Woggon, Ulrike K.; Budde, Ina; Malzacher, Mario; Pufahl, Karsten
    The textile industry has a detrimental impact on the environment and faces challenges to meet climate and environmental goals. The core problem is its linear system where only a small fraction of used garments is reused or recycled. To facilitate recycling of textiles, a sorting method is proposed that provides the necessary information about a garment to sort it into the appropriate recycling channel. Current methods have difficulties identifying garments polluted by chemicals used during dyeing and finishing which hinder recycling into new high-quality fibers. In this concept paper, the Circular Textile Intelligence (CRTX) project is outlined which addresses the central problem of linear structures in the textile industry. A spectroscopy and AI-based method for textile sorting will enable more informed decisions about optimal reuse and recycling channels, thus closing the loop. It will build upon a material database containing information about materials, textiles, dyes, finishings, and unwanted hazardous substances.
  • Konferenzbeitrag
    Predicting How to Test Requirements: An Automated Approach
    (Software Engineering 2020, 2020) Winkler, Jonas; Grönberg, Jannis; Vogelsang, Andreas
    An important task in requirements engineering is to identify and determine how to verify a requirement (eg., by manual review, testing, or simulation; also called \emphpotential verification method). This information is required to effectively create test cases and verification plans for requirements. In this paper, we propose an automatic approach to classify natural language requirements with respect to their potential verification methods (PVM). Our approach uses a convolutional neural network architecture to implement a multiclass and multilabel classifier that assigns probabilities to a predefined set of six possible verification methods, which we derived from an industrial guideline. Additionally, we implemented a backtracing approach to analyze and visualize the reasons for the network's decisions. In a 10-fold cross validation on a set of about 27,000 industrial requirements, our approach achieved a macro averaged \fone score of 0.79 across all labels. The results show that our approach might help to increase the quality of requirements specifications with respect to the PVM attribute and guide engineers in effectively deriving test cases and verification plans.