006 Spezielle Computerverfahren
Refine
Departments, institutes and facilities
- Fachbereich Informatik (53)
- Institute of Visual Computing (IVC) (26)
- Fachbereich Wirtschaftswissenschaften (17)
- Institut für Verbraucherinformatik (IVI) (17)
- Institut für Technik, Ressourcenschonung und Energieeffizienz (TREE) (11)
- Institut für Sicherheitsforschung (ISF) (6)
- Fachbereich Ingenieurwissenschaften und Kommunikation (4)
- Graduierteninstitut (3)
- Zentrum für Ethik und Verantwortung (ZEV) (2)
- Fachbereich Angewandte Naturwissenschaften (1)
Document Type
- Conference Object (40)
- Article (35)
- Part of a Book (7)
- Preprint (5)
- Contribution to a Periodical (4)
- Doctoral Thesis (4)
- Report (4)
- Book (monograph, edited volume) (2)
- Research Data (2)
- Patent (1)
Year of publication
Keywords
- Augmented Reality (5)
- Knowledge Graphs (3)
- Machine Learning (3)
- Machine learning (3)
- Virtual Reality (3)
- deep learning (3)
- facial expression analysis (3)
- haptics (3)
- virtual reality (3)
- 3D user interface (2)
While humans can effortlessly pick a view from multiple streams, automatically choosing the best view is a challenge. Choosing the best view from multi-camera streams poses a problem regarding which objective metrics should be considered. Existing works on view selection lack consensus about which metrics should be considered to select the best view. The literature on view selection describes diverse possible metrics. And strategies such as information-theoretic, instructional design, or aesthetics-motivated fail to incorporate all approaches. In this work, we postulate a strategy incorporating information-theoretic and instructional design-based objective metrics to select the best view from a set of views. Traditionally, information-theoretic measures have been used to find the goodness of a view, such as in 3D rendering. We adapted a similar measure known as the viewpoint entropy for real-world 2D images. Additionally, we incorporated similarity penalization to get a more accurate measure of the entropy of a view, which is one of the metrics for the best view selection. Since the choice of the best view is domain-dependent, we chose demonstration-based training scenarios as our use case. The limitation of our chosen scenarios is that they do not include collaborative training and solely feature a single trainer. To incorporate instructional design considerations, we included the trainer’s body pose, face, face when instructing, and hands visibility as metrics. To incorporate domain knowledge we included predetermined regions’ visibility as another metric. All of those metrics are taken into account to produce a parameterized view recommendation approach for demonstration-based training. An online study using recorded multi-camera video streams from a simulation environment was used to validate those metrics. Furthermore, the responses from the online study were used to optimize the view recommendation performance with a normalized discounted cumulative gain (NDCG) value of 0.912, which shows good performance with respect to matching user choices.
This paper addresses the classification of Arabic text data in the field of Natural Language Processing (NLP), with a particular focus on Natural Language Inference (NLI) and Contradiction Detection (CD). Arabic is considered a resource-poor language, meaning that there are few data sets available, which leads to limited availability of NLP methods. To overcome this limitation, we create a dedicated data set from publicly available resources. Subsequently, transformer-based machine learning models are being trained and evaluated. We find that a language-specific model (AraBERT) performs competitively with state-of-the-art multilingual approaches, when we apply linguistically informed pre-training methods such as Named Entity Recognition (NER). To our knowledge, this is the first large-scale evaluation for this task in Arabic, as well as the first application of multi-task pre-training in this context.
This research investigates the efficacy of multisensory cues for locating targets in Augmented Reality (AR). Sensory constraints can impair perception and attention in AR, leading to reduced performance due to factors such as conflicting visual cues or a restricted field of view. To address these limitations, the research proposes head-based multisensory guidance methods that leverage audio-tactile cues to direct users' attention towards target locations. The research findings demonstrate that this approach can effectively reduce the influence of sensory constraints, resulting in improved search performance in AR. Additionally, the thesis discusses the limitations of the proposed methods and provides recommendations for future research.
The perceptual upright results from the multisensory integration of the directions indicated by vision and gravity as well as a prior assumption that upright is towards the head. The direction of gravity is signalled by multiple cues, the predominant of which are the otoliths of the vestibular system and somatosensory information from contact with the support surface. Here, we used neutral buoyancy to remove somatosensory information while retaining vestibular cues, thus "splitting the gravity vector" leaving only the vestibular component. In this way, neutral buoyancy can be used as a microgravity analogue. We assessed spatial orientation using the oriented character recognition test (OChaRT, which yields the perceptual upright, PU) under both neutrally buoyant and terrestrial conditions. The effect of visual cues to upright (the visual effect) was reduced under neutral buoyancy compared to on land but the influence of gravity was unaffected. We found no significant change in the relative weighting of vision, gravity, or body cues, in contrast to results found both in long-duration microgravity and during head-down bed rest. These results indicate a relatively minor role for somatosensation in determining the perceptual upright in the presence of vestibular cues. Short-duration neutral buoyancy is a weak analogue for microgravity exposure in terms of its perceptual consequences compared to long-duration head-down bed rest.
Neutral buoyancy has been used as an analog for microgravity from the earliest days of human spaceflight. Compared to other options on Earth, neutral buoyancy is relatively inexpensive and presents little danger to astronauts while simulating some aspects of microgravity. Neutral buoyancy removes somatosensory cues to the direction of gravity but leaves vestibular cues intact. Removal of both somatosensory and direction of gravity cues while floating in microgravity or using virtual reality to establish conflicts between them has been shown to affect the perception of distance traveled in response to visual motion (vection) and the perception of distance. Does removal of somatosensory cues alone by neutral buoyancy similarly impact these perceptions? During neutral buoyancy we found no significant difference in either perceived distance traveled nor perceived size relative to Earth-normal conditions. This contrasts with differences in linear vection reported between short- and long-duration microgravity and Earth-normal conditions. These results indicate that neutral buoyancy is not an effective analog for microgravity for these perceptual effects.
The increasing ubiquity of Artificial Intelligence (AI) poses significant political consequences. The rapid proliferation of AI over the past decade has prompted legislators and regulators to attempt to contain AI’s technological consequences. For Germany, relevant design requirements have been expressed by the European Commission’s High-Level Expert Group on Artificial Intelligence (HLEG AI), and, at the national level, by the German government’s Data Ethics Commission (DEK) as well as the German Bundestag’s Commission of Inquiry on Artificial Intelligence (EKKI).
Wie KI Innere Führung lernt
(2022)
Dass sich künstliche Intelligenz (KI) weltweit ausgebreitet hat, ist eine Binsenwahrheit. Die rasche und unaufhaltsame Proliferation von KI der letzten zehn Jahre spricht für sich, und längst ziehen auch Gesetzgeber und Regulierungsbehörden nach, um KI und ihre Technikfolgen einzuhegen. Für Deutschland relevante Gestaltungsanforderungen haben die High-Level Expert Group on Artificial Intelligence der Europäischen Kommission (HLEG AI) und auf nationaler Ebene die Datenethikkommission der Bundesregierung (DEK) und die Enquetekommission Künstliche Intelligenz des Deutschen Bundestags (EKKI) geäußert.
In recent years, the ability of intelligent systems to be understood by developers and users has received growing attention. This holds in particular for social robots, which are supposed to act autonomously in the vicinity of human users and are known to raise peculiar, often unrealistic attributions and expectations. However, explainable models that, on the one hand, allow a robot to generate lively and autonomous behavior and, on the other, enable it to provide human-compatible explanations for this behavior are missing. In order to develop such a self-explaining autonomous social robot, we have equipped a robot with own needs that autonomously trigger intentions and proactive behavior, and form the basis for understandable self-explanations. Previous research has shown that undesirable robot behavior is rated more positively after receiving an explanation. We thus aim to equip a social robot with the capability to automatically generate verbal explanations of its own behavior, by tracing its internal decision-making routes. The goal is to generate social robot behavior in a way that is generally interpretable, and therefore explainable on a socio-behavioral level increasing users' understanding of the robot's behavior. In this article, we present a social robot interaction architecture, designed to autonomously generate social behavior and self-explanations. We set out requirements for explainable behavior generation architectures and propose a socio-interactive framework for behavior explanations in social human-robot interactions that enables explaining and elaborating according to users' needs for explanation that emerge within an interaction. Consequently, we introduce an interactive explanation dialog flow concept that incorporates empirically validated explanation types. These concepts are realized within the interaction architecture of a social robot, and integrated with its dialog processing modules. We present the components of this interaction architecture and explain their integration to autonomously generate social behaviors as well as verbal self-explanations. Lastly, we report results from a qualitative evaluation of a working prototype in a laboratory setting, showing that (1) the robot is able to autonomously generate naturalistic social behavior, and (2) the robot is able to verbally self-explain its behavior to the user in line with users' requests.
Introduction. The experience of pain is regularly accompanied by facial expressions. The gold standard for analyzing these facial expressions is the Facial Action Coding System (FACS), which provides so-called action units (AUs) as parametrical indicators of facial muscular activity. Particular combinations of AUs have appeared to be pain-indicative. The manual coding of AUs is, however, too time- and labor-intensive in clinical practice. New developments in automatic facial expression analysis have promised to enable automatic detection of AUs, which might be used for pain detection. Objective. Our aim is to compare manual with automatic AU coding of facial expressions of pain. Methods. FaceReader7 was used for automatic AU detection. We compared the performance of FaceReader7 using videos of 40 participants (20 younger with a mean age of 25.7 years and 20 older with a mean age of 52.1 years) undergoing experimentally induced heat pain to manually coded AUs as gold standard labeling. Percentages of correctly and falsely classified AUs were calculated, and we computed as indicators of congruency, "sensitivity/recall," "precision," and "overall agreement (F1)." Results. The automatic coding of AUs only showed poor to moderate outcomes regarding sensitivity/recall, precision, and F1. The congruency was better for younger compared to older faces and was better for pain-indicative AUs compared to other AUs. Conclusion. At the moment, automatic analyses of genuine facial expressions of pain may qualify at best as semiautomatic systems, which require further validation by human observers before they can be used to validly assess facial expressions of pain.
Dieses Dokument präsentiert eine Zusammenfassung der Dissertation der Autorin. In dieser Dissertation [Ha20] wurde ein neuartiger und hybrider Ansatz für die Scha ̈tzung der Intensität von Gesichtsmuskelbewegungen (Action Unit (AU)) vorgeschlagen und validiert. Dieser Ansatz basiert auf einer Gauß’schen Zustandsschätzung und kombiniert ein verformbares, AU-basiertes Gesichtsformmodell, ein viskoelastisches Modell der Gesichtsmuskelbewegung, mehrere erscheinungsbasierten AU-Klassifikatoren und eine Methode zur Erkennung von Gesichtspunkten. Es wurden mehrere Erweiterungen vorgeschlagen und in das Zustandsschätzungs-Framework integriert, um mit den personenspezifischen Eigenschaften sowie technischen und praktischen Herausforderungen umzugehen.Die mit der vorgeschlagenen Methode erzeugten AU-Intensitätsschätzungen wurden für die automatische Erkennung von Schmerzen und für die Analyse von Fahrerablenkung eingesetzt.
This dissertation presents a probabilistic state estimation framework for integrating data-driven machine learning models and a deformable facial shape model in order to estimate continuous-valued intensities of 22 different facial muscle movements, known as Action Units (AU), defined in the Facial Action Coding System (FACS). A practical approach is proposed and validated for integrating class-wise probability scores from machine learning models within a Gaussian state estimation framework. Furthermore, driven mass-spring-damper models are applied for modelling the dynamics of facial muscle movements. Both facial shape and appearance information are used for estimating AU intensities, making it a hybrid approach. Several features are designed and explored to help the probabilistic framework to deal with multiple challenges involved in automatic AU detection. The proposed AU intensity estimation method and its features are evaluated quantitatively and qualitatively using three different datasets containing either spontaneous or acted facial expressions with AU annotations. The proposed method produced temporally smoother estimates that facilitate a fine-grained analysis of facial expressions. It also performed reasonably well, even though it simultaneously estimates intensities of 22 AUs, some of which are subtle in expression or resemble each other closely. The estimated AU intensities tended to the lower range of values, and were often accompanied by a small delay in onset. This shows that the proposed method is conservative. In order to further improve performance, state-of-the-art machine learning approaches for AU detection could be integrated within the proposed probabilistic AU intensity estimation framework.
It is only a matter of time until autonomous vehicles become ubiquitous; however, human driving supervision will remain a necessity for decades. To assess the drive's ability to take control over the vehicle in critical scenarios, driver distractions can be monitored using wearable sensors or sensors that are embedded in the vehicle, such as video cameras. The types of driving distractions that can be sensed with various sensors is an open research question that this study attempts to answer. This study compared data from physiological sensors (palm electrodermal activity (pEDA), heart rate and breathing rate) and visual sensors (eye tracking, pupil diameter, nasal EDA (nEDA), emotional activation and facial action units (AUs)) for the detection of four types of distractions. The dataset was collected in a previous driving simulation study. The statistical tests showed that the most informative feature/modality for detecting driver distraction depends on the type of distraction, with emotional activation and AUs being the most promising. The experimental comparison of seven classical machine learning (ML) and seven end-to-end deep learning (DL) methods, which were evaluated on a separate test set of 10 subjects, showed that when classifying windows into distracted or not distracted, the highest F1-score of 79%; was realized by the extreme gradient boosting (XGB) classifier using 60-second windows of AUs as input. When classifying complete driving sessions, XGB's F1-score was 94%. The best-performing DL model was a spectro-temporal ResNet, which realized an F1-score of 75%; when classifying segments and an F1-score of 87%; when classifying complete driving sessions. Finally, this study identified and discussed problems, such as label jitter, scenario overfitting and unsatisfactory generalization performance, that may adversely affect related ML approaches.
Towards an Interaction-Centered and Dynamically Constructed Episodic Memory for Social Robots
(2020)
Towards self-explaining social robots. Verbal explanation strategies for a needs-based architecture
(2019)
In order to establish long-term relationships with users, social companion robots and their behaviors need to be comprehensible. Purely reactive behavior such as answering questions or following commands can be readily interpreted by users. However, the robot's proactive behaviors, included in order to increase liveliness and improve the user experience, often raise a need for explanation. In this paper, we provide a concept to produce accessible “why-explanations” for the goal-directed behavior an autonomous, lively robot might produce. To this end we present an architecture that provides reasons for behaviors in terms of comprehensible needs and strategies of the robot, and we propose a model for generating different kinds of explanations.
Towards explaining deep learning networks to distinguish facial expressions of pain and emotions
(2018)
Deep learning networks are successfully used for object and face recognition in images and videos. In order to be able to apply such networks in practice, for example in hospitals as a pain recognition tool, the current procedures are only suitable to a limited extent. The advantage of deep learning methods is that they can learn complex non-linear relationships between raw data and target classes without limiting themselves to a set of hand-crafted features provided by humans. However, the disadvantage is that due to the complexity of these networks, it is not possible to interpret the knowledge that is stored inside the network. It is a black-box learning procedure. Explainable Artificial Intelligence (AI) approaches mitigate this problem by extracting explanations for decisions and representing them in a human-interpretable form. The aim of this paper is to investigate the explainable AI method Layer-wise Relevance Propagation (LRP) and apply it to explain how a deep learning network distinguishes facial expressions of pain from facial expressions of emotions such as happiness and disgust.
This paper describes a dynamic, model-based approach for estimating intensities of 22 out of 44 different basic facial muscle movements. These movements are defined as Action Units (AU) in the Facial Action Coding System (FACS) [1]. The maximum facial shape deformations that can be caused by the 22 AUs are represented as vectors in an anatomically based, deformable, point-based face model. The amount of deformation along these vectors represent the AU intensities, and its valid range is [0, 1]. An Extended Kalman Filter (EKF) with state constraints is used to estimate the AU intensities. The focus of this paper is on the modeling of constraints in order to impose the anatomically valid AU intensity range of [0, 1]. Two process models are considered, namely constant velocity and driven mass-spring-damper. The results show the temporal smoothing and disambiguation effect of the constrained EKF approach, when compared to the frame-by-frame model fitting approach ‘Regularized Landmark Mean-Shift (RLMS)’ [2]. This effect led to more than 35% increase in performance on a database of posed facial expressions.
A device includes an input to sequential data associated to a face; a predictor configured to predict facial parameters; and a corrector configured to correct the predicted facial parameters on the basis of input data, the input data containing geometric measurements and other information. A related method and a related computer program are also disclosed.
BACKGROUND
Given the unreliable self-report in patients with dementia, pain assessment should also rely on the observation of pain behaviors, such as facial expressions. Ideal observers should be well trained and should observe the patient continuously in order to pick up any pain-indicative behavior; which are requisitions beyond realistic possibilities of pain care. Therefore, the need for video-based pain detection systems has been repeatedly voiced. Such systems would allow for constant monitoring of pain behaviors and thereby allow for a timely adjustment of pain management in these fragile patients, who are often undertreated for pain.
METHODS
In this road map paper we describe an interdisciplinary approach to develop such a video-based pain detection system. The development starts with the selection of appropriate video material of people in pain as well as the development of technical methods to capture their faces. Furthermore, single facial motions are automatically extracted according to an international coding system. Computer algorithms are trained to detect the combination and timing of those motions, which are pain-indicative.
RESULTS/CONCLUSION
We hope to encourage colleagues to join forces and to inform end-users about an imminent solution of a pressing pain-care problem. For the near future, implementation of such systems can be foreseen to monitor immobile patients in intensive and postoperative care situations.
A method for minimum range extension with improved accuracy in triangulation laser range finder
(2011)
Robust Identification and Segmentation of the Outer Skin Layers in Volumetric Fingerprint Data
(2022)
Despite the long history of fingerprint biometrics and its use to authenticate individuals, there are still some unsolved challenges with fingerprint acquisition and presentation attack detection (PAD). Currently available commercial fingerprint capture devices struggle with non-ideal skin conditions, including soft skin in infants. They are also susceptible to presentation attacks, which limits their applicability in unsupervised scenarios such as border control. Optical coherence tomography (OCT) could be a promising solution to these problems. In this work, we propose a digital signal processing chain for segmenting two complementary fingerprints from the same OCT fingertip scan: One fingerprint is captured as usual from the epidermis (“outer fingerprint”), whereas the other is taken from inside the skin, at the junction between the epidermis and the underlying dermis (“inner fingerprint”). The resulting 3D fingerprints are then converted to a conventional 2D grayscale representation from which minutiae points can be extracted using existing methods. Our approach is device-independent and has been proven to work with two different time domain OCT scanners. Using efficient GPGPU computing, it took less than a second to process an entire gigabyte of OCT data. To validate the results, we captured OCT fingerprints of 130 individual fingers and compared them with conventional 2D fingerprints of the same fingers. We found that both the outer and inner OCT fingerprints were backward compatible with conventional 2D fingerprints, with the inner fingerprint generally being less damaged and, therefore, more reliable.
The visual and auditory quality of computer-mediated stimuli for virtual and extended reality (VR/XR) is rapidly improving. Still, it remains challenging to provide a fully embodied sensation and awareness of objects surrounding, approaching, or touching us in a 3D environment, though it can greatly aid task performance in a 3D user interface. For example, feedback can provide warning signals for potential collisions (e.g., bumping into an obstacle while navigating) or pinpointing areas where one’s attention should be directed to (e.g., points of interest or danger). These events inform our motor behaviour and are often associated with perception mechanisms associated with our so-called peripersonal and extrapersonal space models that relate our body to object distance, direction, and contact point/impact. We will discuss these references spaces to explain the role of different cues in our motor action responses that underlie 3D interaction tasks. However, providing proximity and collision cues can be challenging. Various full-body vibration systems have been developed that stimulate body parts other than the hands, but can have limitations in their applicability and feasibility due to their cost and effort to operate, as well as hygienic considerations associated with e.g., Covid-19. Informed by results of a prior study using low-frequencies for collision feedback, in this paper we look at an unobtrusive way to provide spatial, proximal and collision cues. Specifically, we assess the potential of foot sole stimulation to provide cues about object direction and relative distance, as well as collision direction and force of impact. Results indicate that in particular vibration-based stimuli could be useful within the frame of peripersonal and extrapersonal space perception that support 3DUI tasks. Current results favor the feedback combination of continuous vibrotactor cues for proximity, and bass-shaker cues for body collision. Results show that users could rather easily judge the different cues at a reasonably high granularity. This granularity may be sufficient to support common navigation tasks in a 3DUI.
Taste is a complex phenomenon that depends on the individual experience and is a matter of collective negotiation and mediation. On the contrary, it is uncommon to include taste and its many facets in everyday design, particularly online shopping for fresh food products. To realize this unused potential, we conducted two Co-Design workshops. Based on the participants’ results in the workshops, we prototyped and evaluated a click-dummy smart-phone app to explore consumers’ needs for digital taste depiction. We found that emphasizing the natural qualities of food products, external reviews, and personalizing features lead to a reflection on the individual taste experience. The self-reflection through our design enables consumers to develop their taste competencies and thus strengthen their autonomy in decision-making. Ultimately, exploring taste as a social experience adds to a broader understanding of taste beyond a sensory phenomenon.
We describe a systematic approach for rendering time-varying simulation data produced by exa-scale simulations, using GPU workstations. The data sets we focus on use adaptive mesh refinement (AMR) to overcome memory bandwidth limitations by representing interesting regions in space with high detail. Particularly, our focus is on data sets where the AMR hierarchy is fixed and does not change over time. Our study is motivated by the NASA Exajet, a large computational fluid dynamics simulation of a civilian cargo aircraft that consists of 423 simulation time steps, each storing 2.5 GB of data per scalar field, amounting to a total of 4 TB. We present strategies for rendering this time series data set with smooth animation and at interactive rates using current generation GPUs. We start with an unoptimized baseline and step by step extend that to support fast streaming updates. Our approach demonstrates how to push current visualization workstations and modern visualization APIs to their limits to achieve interactive visualization of exa-scale time series data sets.
This paper explores the role of artificial intelligence (AI) in elite sports. We approach the topic from two perspectives. Firstly, we provide a literature based overview of AI success stories in areas other than sports. We identified multiple approaches in the area of Machine Perception, Machine Learning and Modeling, Planning and Optimization as well as Interaction and Intervention, holding a potential for improving training and competition. Secondly, we discover the present status of AI use in elite sports. Therefore, in addition to another literature review, we interviewed leading sports scientist, which are closely connected to the main national service institute for elite sports in their countries. The analysis of this literature review and the interviews show that the most activity is carried out in the methodical categories of signal and image processing. However, projects in the field of modeling & planning have become increasingly popular within the last years. Based on these two perspectives, we extract deficits, issues and opportunities and summarize them in six key challenges faced by the sports analytics community. These challenges include data collection, controllability of an AI by the practitioners and explainability of AI results.
The latest trends in inverse rendering techniques for reconstruction use neural networks to learn 3D representations as neural fields. NeRF-based techniques fit multi-layer perceptrons (MLPs) to a set of training images to estimate a radiance field which can then be rendered from any virtual camera by means of volume rendering algorithms. Major drawbacks of these representations are the lack of well-defined surfaces and non-interactive rendering times, as wide and deep MLPs must be queried millions of times per single frame. These limitations have recently been singularly overcome, but managing to accomplish this simultaneously opens up new use cases. We present KiloNeuS, a new neural object representation that can be rendered in path-traced scenes at interactive frame rates. KiloNeuS enables the simulation of realistic light interactions between neural and classic primitives in shared scenes, and it demonstrably performs in real-time with plenty of room for future optimizations and extensions.
ProtSTonKGs: A Sophisticated Transformer Trained on Protein Sequences, Text, and Knowledge Graphs
(2022)
While most approaches individually exploit unstructured data from the biomedical literature or structured data from biomedical knowledge graphs, their union can better exploit the advantages of such approaches, ultimately improving representations of biology. Using multimodal transformers for such purposes can improve performance on context dependent classication tasks, as demonstrated by our previous model, the Sophisticated Transformer Trained on Biomedical Text and Knowledge Graphs (STonKGs). In this work, we introduce ProtSTonKGs, a transformer aimed at learning all-encompassing representations of protein-protein interactions. ProtSTonKGs presents an extension to our previous work by adding textual protein descriptions and amino acid sequences (i.e., structural information) to the text- and knowledge graph-based input sequence used in STonKGs. We benchmark ProtSTonKGs against STonKGs, resulting in improved F1 scores by up to 0.066 (i.e., from 0.204 to 0.270) in several tasks such as predicting protein interactions in several contexts. Our work demonstrates how multimodal transformers can be used to integrate heterogeneous sources of information, paving the foundation for future approaches that use multiple modalities for biomedical applications.
Vection underwater
(2022)