Skip directly to content

Coming theses from other universities

Please note that the date and time given on these pages is the time of electronic publication, and not the date of the public defense. To find the time and venue of the public defense, please follow the link to DiVA of the thesis in question.
  • A Socio-Material Stuy of User Involvement : Interrogating the practices of technology development for older people in a digitalised world

    Author: Björn Fischer
    Publication date: 2022-06-22 14:48

    Population ageing and increased digitalization each constitute an ongoing and profound transformation within contemporary modes of living, as growing advances in technological development mix and intermingle with the lived realities of older people as the final recipients. It is against the backdrop of this interplay that user involvement has enjoyed ever-rising advocacy to an almost normative degree. Beyond articulating methodological principles, however, the literature has remained surprisingly vague as to the practical implementation of the approach. Less appears to be known, both empirically and conceptually, about how design and user involvement are done in practice and how they would matter to bring about intentional or unintentional effects. 

    To engage with these developments, this thesis aims at taking the practices of user involvement and design to the centre of its inquiry by adopting a perspective from Science and Technology Studies (STS). Specifically, the thesis seeks to both build on and contribute to the established body of STS on the connection between technology design and older users and ask: What is there to learn about user involvement as a method, if we focus on the practices of doing user involvement? To answer this question, the thesis studies four different aspects of the practices of user involvement and design. In particular, the thesis reviews the literature on how user involvement mattered in previous empirical projects that include older people (Paper I), it examines how different configurations of participation matter in design workshops (Paper II), it scrutinizes the achievement of user involvement in corporate practices (Paper III) and it traces the circumstantial performances of such practices (Paper IV). The largest empirical piece comes from a two-year ethnographic study of a small- to medium-sized enterprise, the material from which informed Paper III and Paper IV.

    The findings highlight how user involvement in practice is both contingent and transformative, as it selectively enrols different participants and performs multiple realities. In practice, user involvement appears to be dependent on a set of underlying premises and socio-material conditions and thus is always a dynamic and momentary achievement. Furthermore, the thesis shows how the practices of user involvement themselves may bring into existence different realities, articulating and materializing particular versions of objects and images of ageing. Accordingly, the thesis contributes theoretically by illuminating the underlying socio-material facets of user involvement, and by emphasizing ageing as a particular object/image of design. Specifically, the appended papers encompass a conceptual framework, as well as three new concepts: design multiple, shifting interstices and viscous image landscape, in order to theorize the underlying conditions for user involvement, its relationship with design and its entanglement with ageing. Practically, the thesis enunciates three main implications regarding questions of goodness, politics and ethics.

  • Learning Spatiotemporal Features in Low-Data and Fine-Grained Action Recognition with an Application to Equine Pain Behavior

    Author: Sofia Broomé
    Publication date: 2022-06-16 15:32

    Recognition of pain in animals is important because pain compromises animal welfare and can be a manifestation of disease. This is a difficult task for veterinarians and caretakers, partly because horses, being prey animals, display subtle pain behavior, and because they cannot verbalize their pain. An automated video-based system has a large potential to improve the consistency and efficiency of pain predictions.

    Video recording is desirable for ethological studies because it interferes minimally with the animal, in contrast to more invasive measurement techniques, such as accelerometers. Moreover, to be able to say something meaningful about animal behavior, the subject needs to be studied for longer than the exposure of single images. In deep learning, we have not come as far for video as we have for single images, and even more questions remain regarding what types of architectures should be used and what these models are actually learning. Collecting video data with controlled moderate pain labels is both laborious and involves real animals, and the amount of such data should therefore be limited. The low-data scenario, in particular, is under-explored in action recognition, in favor of the ongoing exploration of how well large models can learn large datasets.

    The first theme of the thesis is automated recognition of equine pain. Here, we propose a method for end-to-end equine pain recognition from video, finding, in particular, that the temporal modeling ability of the artificial neural network is important to improve the classification. We surpass veterinarian experts on a dataset with horses undergoing well-defined moderate experimental pain induction.  Next, we investigate domain transfer to another type of pain in horses: less defined, longer-acting and lower-grade orthopedic pain. We find that a smaller, recurrent video model is more robust to domain shift on a target dataset than a large, pre-trained, 3D CNN, having equal performance on a source dataset. We also discuss challenges with learning video features on real-world datasets.

    Motivated by questions arisen within the application area, the second theme of the thesis is empirical properties of deep video models. Here, we study the spatiotemporal features that are learned by deep video models in end-to-end video classification and propose an explainability method as a tool for such investigations. Further, the question of whether different approaches to frame dependency treatment in video models affect their cross-domain generalization ability is explored through empirical study. We also propose new datasets for light-weight temporal modeling and to investigate texture bias within action recognition.

  • Optical Performance Monitoring in Digital Coherent Communications: Intelligent Error Vector Magnitude Estimation

    Author: Yuchuan Fan
    Publication date: 2022-06-15 08:03

    The rapid development of data-driven techniques brings us new applications, such asfifth-generation new radio (5G NR), high-definition video, Internet of things (IoT),etc., which has greatly facilitated our daily lives. Optical networks as one fundamen-tal infrastructure are evolving to simultaneously support these high dimensional dataservices, with a feature of flexible, dynamic, and heterogeneous. Optical performancemonitoring (OPM) is a key enabler to guarantee reliable network management andmaintenance, which improving network controllability and resource efficiency. Accu-rately telemetry key performance indicators (KPIs) such as bit error rate (BER) canextend monitoring functionality and secure network management. However, retrievingthe BER level metric is time-consuming and inconvenient for OPM. Low-complexityOPM strategies are highly desired for ubiquitous departments at optical network nodes.This thesis investigates machine learning (ML) based intelligent error vector mag-nitude (EVM) estimation schemes in digital coherent communications, where EVMis widely used as an alternative BER metric for multilevel modulated signals. Wepropose a prototype of EVM estimation, which enables monitoring signal quality froma short observation period. Three alternative ML algorithms are explored to facilitatethe implementation of this prototype, namely convolutional neural networks (CNNs),feedforward neural networks (FFNNs), and linear regression (LR). We show that CNNconjunction with graphical signal representations, i.e., constellation diagrams and am-plitude histograms (AHs), can achieve decent EVM estimation accuracy for signalsbefore and after carrier phase recovery (CPR), which outperforms the conventionalEVM calculation. Moreover, we show that an FFNN-based scheme can reduce poten-tial energy and keep the estimation accuracy by directly operating with AH vectors.Furthermore, the estimation capability is thoroughly studied when the system hasdifferent impairments. Lastly, we demonstrate that a simple LR-designed model canperform as well as FFNN when the information on modulation formats is known. SuchLR-based can be easily implemented with modulation formats identification modulein OPM, providing accurate signal quality information.

  • Transparent but incomprehensible : Investigating the relation between transparency, explanations,and usability in automated decision-making

    Author: Jacob Dexe
    Publication date: 2022-06-13 09:20

    Transparency is almost always seen as a desirable state of affairs. Governments should be more transparent towards their citizens, and corporations should be more transparent towards both public authorities and their customers. More transparency means more information which citizens can use to make decisions about their daily lives, and with increasing amounts of information in society, those citizens would be able to make more and more choices that align with their preferences. It is just that the story is slightly too good to be true. Instead, citizens are skeptical towards increased data collection, demand harsher transparency requirements and seem to lack both time and ability to properly engage with all the information available.

    In this thesis the relation between transparency, explanations and usability is investigated within the context of automated decision-making. Aside from showing the benefits that transparency can have, it shows a wide array of different problems with transparency, and how transparency can be harder to accomplish than most assume. This thesis explores the explanations, which often make up the transparency, and their limitations, developments in automation and algorithmic decisions, as well as how society tends to regulate such things. It then applies these frameworks and investigates how human-computer interaction in general, and usability in particular, can help improve how transparency can bring the many benefits it promises.

    Four papers are presented that study the topic from various perspectives. Paper I looks at how governments give guidance in achieving competitive advantages with ethical AI, while Paper II studies how insurance professionals view the benefits and limitations of transparency. Paper III and IV both study transparency in practice by use of requests for information according to GDPR. But while Paper III provides a comparative study of GDPR implementation in five countries, Paper IV instead shows and explores how transparency can fail and ponders why.

    The thesis concludes by showing that while transparency does indeed have many benefits, it also has limitations. Companies and other actors need to be aware that sometimes transparency is simply not the right solution, and explanations have limitations for both automation and in humans. Transparency as a tool can reach certain goals, but good transparency requires good strategies, active choices and an awareness of what users need.

  • Final-State-Resolved Mutual Neutralization of Li+ and H-

    Author: Alice F. Schmidt-May
    Publication date: 2022-06-07 15:36

    We studied the mutual neutralization of Li+ and H- at effective collision energies of a few hundred meV, which corresponds to temperatures of around 2000 K, in the double ion storage ring DESIREE.We present a new approach to match beam velocities and a new general analysis method for non-fragmenting mutual neutralization at DESIREE.Our results show two features, which we could clearly assign to the product channel into the electronically   excited  3s state of neutral lithium and an unresolved combination of 3p and 3d final state contributions.Branching fractions into 3s are extracted for ten different collision energies via spectral binning and compared to several theoretical investigations and two previous measurements, which focused on the heavier isotope deuterium.We find a significant isotope effect, as theoretically predicted, but in contrast to previous experimental results. The branching fractions agree well with different theoretical approaches using non-empirical couplings and  best with a combination of ab initio potentials and Landau-Zener transition probabilities.

  • Intelligent System Designs : Data-driven Sensor Calibration & Smart Meter Privacy

    Author: Yang You
    Publication date: 2022-06-02 16:17

    Nowadays, the intelligent system has gained high popularity in successful implementation of real-time tasks due to its capability of providing efficient and powerful decision making in real applications. In this thesis, we aim for exploring and exploiting different concepts or methods to handle different tasks towards the intelligent system design. In particular, we focus on the following two problems: (i) Consumer-centric privacy-cost trade-off in smart metering system; (ii) Data-driven calibration for gas sensing system.

    For the first target problem, an optimal privacy-preserving and cost-efficient energy management strategy is designed for each smart grid consumer that is equipped with a rechargeable energy storage. The Kullback-Leibler divergence rate is used as privacy measure and the expected cost-saving rate is used as utility measure. The corresponding energy management strategy is designed by optimizing a weighted sum of both privacy and cost measures over a finite time horizon, which is achieved by  formulating our problem into a partial observed Markov decision process problem. A computationally efficient approximated Q-learning method is proposed as a extension to high-dimensional problems over an infinite time horizon. 

    Furthermore, the privacy-preserving and cost-efficient energy management strategy is designed for multiple smart grid consumers that are equipped with renewable energy sources. Different from the previous problem, the adversary is assumed to employ a factorial hidden Markov model based inference for load disaggregation, and the corresponding joint log-likelihood of the model is utilized as privacy measure. A dynamic pricing model is studied, where the price of unit amount of energy is determined by the consumers' aggregated power request, which suits a commodity-limited market. The consumers' energy management strategy is designed under a non-cooperative game framework by optimizing a weighted sum of both privacy measure and the user's energy cost savings. The consumers' non-cooperative game is shown to admit a unique pure strategy Nash equilibrium. As an extension, a computational-efficient distributed Nash equilibrium energy management strategy seeking method is proposed, which also avoids the privacy leakage due to the sharing of payoff functions between consumers.

    For the second target problem, several data-driven self-calibration algorithms are developed for low-cost non-dispersive infrared sensors. The measurement errors of the sensors are mainly caused by the remaining model errors and can be fully described by the drift of the calibration parameter. This leads to our first formulation of a statistical inference problem on the true calibration parameter under the HMM framework, which is a stochastic model that jointly builds on different quantities introduced by the physical model. To better track the time-varying drift process of the sensor, a time-adaptive expectation maximization learning framework is proposed to efficiently update the HMM parameters. For the joint calibration of the gas sensing system, sensors firsttransmit their belief functions of the true gas concentration levelto the cloud. Then the cloud fusion center computes a fusedbelief function according to certain rules. This belief functionis then used as reference for calibrating the sensors. To dealwith the case where belief functions highly conflict with eachother, a Wasserstein distance based weighted average belieffunction fusion approach is first proposed as networked calibration algorithm. To achieve more long-term stable calibration results, the networked calibration problem is further formulated as a partially observed Markov decision process problem, and the calibration strategies are derived in a sequential manner. Correspondingly, the deep Q-network approach is applied as a computationally efficient method to solve the proposed Markovdecision process problem.

    The results in this thesis have shown that our proposed design frameworks can provide concise but precise mathematical models, proper problem formulations, and efficient solutions for the target design objectives of different intelligent systems. 

  • Functions and memory features of adaptive- and innate immune cells in physiological and inflammatory settings

    Author: Gintare Lasaviciute
    Publication date: 2022-06-01 09:00

    The immune system is a complex but well-regulated network which cooperates with the microbiota to maintain optimal homeostasis under physiological settings. A number of factors display the capacity to alter the immune system and thus microbiota crosstalk including host genetics, diet, environmental influences and drugs such as antibiotics or chemotherapeutic agents.

    We and others have previously demonstrated immune regulatory capacities of the gut commensal Limosilactobacillus reuteri (L. reuteri) that seem to act via myeloid cells. In paper I, we have further shown that priming of blood dendritic cells (DCs) or monocytes with L. reuteri-derived cell free supernatant (CFS) modifies how these cells respond to future stimulus challenge, even after monocytes differentiation to DCs (mo-DCs). Notably, priming conditions in mo-DCs skewed subsequent T-helper cell responses. In paper II, we continued to elaborate on whether microbial and gut-associated metabolites modify chemotherapy-induced effects. Thus far, we could show that the CFS from probiotic bacteria might imprint immune cells to modulate inflammatory responses when intestinal epithelial cells are compromised by chemotherapy exposure. In paper III, we investigated how chemotherapy agent Doxorubicin (Doxo) affects bone marrow resident mesenchymal stromal cells (BM MSCs) which support antibody secreting cells (ASCs) responsible for serological memory. Our findings have shown that Doxo-induced alterations in BM MSCs are not sufficient to disrupt their support to ASCs, thus alternative or additional factors might be implicated in ASC preservation during chemotherapy. Lastly, in paper IV, we investigated how Doxo affects secondary lymphoid organs and we found that splenic compartment is more prominently affected by Doxo treatment compared to its lymph node counterpart in an animal model of rhesus macaques.

    Collectively, this thesis outlines novel perceptions on innate immune memory-like capacity and how gut-associated factors influence such recall responses in innate immune cells. Further, this thesis increases our knowledge on how adaptive immune cells, required for long-term serological memory, are preserved during toxic conditions such as those induced by chemotherapy treatment.

  • Bainite Formation Studied by In-situ High-energy X-ray Diffraction

    Author: Sen Lin
    Publication date: 2022-05-31 16:17

          Bainitic steels have attracted great attentions in recent years due to their excellent combination of properties to accommodate a wide range of applications. A deep and comprehensive understanding of how bainite forms is required to better design the production process and optimize the properties of bainitic steels. Extensive experimental investigations, mostly using the post-mortem techniques, have been conducted to shed light on the bainitic transformation. Unfortunately, the nature of bainitic transformation is still a subject of debate, which hinders further development.

          The bainitic transformation involves multiple events that occur concurrently, such as formation of bainitic ferrite and cementite, dislocation generation and annihilation, and carbon diffusion, etc. These events may also affect each other and, in turn, affect the overall bainitic transformation kinetics. It is difficult to quantify the evolution of these events in a large sample volume and reveal their interrelationship during bainitic transformation using conventional experimental methods. Moreover, some bainitic steel grades, such as high-strength low-alloy (HSLA) steels used in the automotive industry, possess a very rapid phase transformation kinetics and a complex final microstructure. So, it is challenging to understand the transformation progress and correlate it with the final microstructure by the sole assistance of post-mortem techniques. In this circumstance, in-situ techniques, such as high-energy X-ray diffraction (HEXRD), appear to be a better solution to these challenges.

          Synchrotron sources provide extremely brilliant X-rays. It enables the detection of minor phases, such as carbides, and facilitates the rapid data acquisition to resolve the rapid transformation progress. Therefore, this thesis is dedicated to utilizing of HEXRD with state-of-the-art instrumentation to study bainite formation. One objective is to explore the feasibility of HEXRD for industrially relevant questions, e.g., rapid bainitic transformation in HSLA steels. At the same time, Si and Mo are two important elements and their content is often tuned in HSLA steels. So, another objective is to systematically study the influence of Si, Mo, and temperature on the bainitic transformation and other events that occur concurrently. The thesis thus brings industrial applications, fundamental transformation mechanisms, and HEXRD methodology development together.

          Two commercial HSLA steels with different hardenabilities were austenitized and fast cooled to different isothermal temperatures. Austenite decomposition occurred during cooling with high transformation rates. Several transformation products, i.e. polygonal ferrite, bainitic ferrite, degenerate pearlite, and martensite, were separated by combining HEXRD and electron backscatter diffraction analyzes. The steel with higher hardenability was found to have a smaller fraction of polygonal ferrite and a higher amount of bainite, which was speculated to be caused by the larger addition of Mo. On the other hand, the low-hardenability steel with a higher addition of Si show a higher carbon content in the retained austenite, probably because of suppressed carbide formation.

          Following the study of HSLA steels, the effects of Si and Mo were investigated using a series of Fe-C-Mn-Si and Fe-C-Mn-Si-Mo alloys with various of heat treatment conditions. These investigations aimed to reveal the influences of the alloying elements with a focus on the correlation between the formation of bainitic ferrite, carbon diffusion, carbide formation, and dislocation density evolution during the bainitic transformation. In general, bainite formation is retarded by increasing the Si content and the isothermal temperature, and the carbon contents in retained austenite at transformation stasis were close to the Widmanstätten/bainitic ferrite start temperatures (WBs). A minor addition of Mo had a negligible effect, but a larger addition introduced a bay area in the time-temperature-transformation that can be a result of solute drag effect. Transition carbides were only found in Si-added alloys, whereas cementite was found in both Si-added and Mo-added alloys. The carbide formation had similar kinetics as bainitic ferrite but no correlation to the dislocation density evolution was found for Si-added alloys.

          Furthermore, an ongoing work is introduced. A HEXRD study using the state-of-the-art PILATUS area detector was performed to shed more light on the transformation mechanism of bainite in comparison with martensite. The result shows that at the same range of diffraction angle, for bainitic ferrite, a single symmetric diffraction peak was found in Fe-3.0Mn-0.4 and 0.6C (in weight percent) alloys; for martensite, the diffraction peak of the Fe-3.0Mn-0.4C alloy was similar to that of bainitic ferrite, whereas two peaks adjacent to each other were found for martensite in the Fe-3Mn-0.6C alloy.

          This thesis demonstrates the versatility of HEXRD in phase transformation studies. Crystallographic, chemical, volume fraction, and stress/strain information extracted from the data is of particular interest for scientific and industrial studies.

  • Multi-Modal Deep Learning with Sentinel-1 and Sentinel-2 Data for Urban Mapping and Change Detection

    Author: Sebastian Hafner
    Publication date: 2022-05-30 12:30

    Driven by the rapid growth in population, urbanization is progressing at an unprecedented rate in many places around the world. Earth observation has become an invaluable tool to monitor urbanization on a global scale by either mapping the extent of cities or detecting newly constructed urban areas within and around cities. In particular, the Sentinel-1 (S1) Synthetic Aperture Radar (SAR) and Sentinel-2 (S2) MultiSpectral Instrument (MSI) missions offer new opportunities for urban mapping and urban Change Detection (CD) due to the capability of systematically acquiring wide-swath high-resolution images with frequent revisits globally.

    Current trends in both urban mapping and urban CD have shifted from employing traditional machine learning methods to Deep Learning (DL) models, specifically Convolutional Neural Networks (CNNs). Recent urban mapping efforts achieved promising results by training CNNs on available built-up data using S2 images. Likewise, DL models have been applied to urban CD problems using S2 data with promising results.

    However, the quality of current methods strongly depends on the availability of local reference data for supervised training, especially since CNNs applied to unseen areas often produce unsatisfactory results due to their insufficient across-region generalization ability. Since multitemporal reference data are even more difficult to obtain, unsupervised learning was suggested for urban CD. While unsupervised models may perform more consistently across different regions, they often perform considerably worse than their supervised counterparts. To alleviate these shortcomings, it is desirable to leverage Semi-Supervised Learning (SSL) that exploits unlabeled data to improve upon supervised learning, especially because satellite data is plentiful. Furthermore, the integration of SAR data into the current optical frameworks (i.e., data fusion) has the potential to produce models with better generalization ability because the representation of urban areas in SAR images is largely invariant across cities, while spectral signatures vary greatly. 

    In this thesis, a novel Domain Adaptation (DA) approach using SSL is first presented. The DA approach jointly exploits Multi-Modal (MM) S1 SAR and S2 MSI to improve across-region generalization for built-up area mapping. Specifically, two identical sub-networks are incorporated into the proposed model to perform built-up area segmentation from SAR and optical images separately. Assuming that consistent built-up area segmentation should be obtained across data modalities, an unsupervised loss for unlabeled data that penalizes inconsistent segmentation from the two sub-networks was designed. Therefore, the use of complementary data modalities as real-world perturbations for Consistency Regularization (CR) is proposed. For the final prediction, the model takes both data modalities into account. Experiments conducted on a test set comprised of sixty representative sites across the world showed that the proposed DA approach achieves strong improvements (F1 score 0.694) upon supervised learning from S1 SAR data (F1 score 0.574), S2 MSI data (F1 score 0.580) and their input-level fusion (F1 score 0.651). The comparison with two state-of-the-art global human settlement maps, namely GHS-S2 and WSF2019, showed that our model is capable of producing built-up area maps with comparable or even better quality.

    For urban CD, a new network architecture for the fusion of SAR and optical data is proposed. Specifically, a dual stream concept was introduced to process different data modalities separately, before combining extracted features at a later decision stage. The individual streams are based on the U-Net architecture. The proposed strategy outperformed other U-Net-based approaches in combination with uni-modal data and MM data with feature level fusion. Furthermore, our approach achieved state-of-the-art performance on the problem posed by a popular urban CD dataset (F1 score 0.600).

    Furthermore, a new network architecture is proposed to adapt Multi-Modal Consistency Regularization (MMCR) for urban CD. Using bi-temporal S1 SAR and S2 MSI image pairs as input, the MM Siamese Difference (Siam-Diff) Dual-Task (DT) network not only predicts changes using a difference decoder, but also segments buildings for each image with a semantic decoder. The proposed network is trained in a semi-supervised fashion using the underlying idea of MMCR, namely that building segmentation across sensor modalities should be consistent, to learn more robust features. The proposed method was tested on an urban CD task using the 60 sites of the SpaceNet7 dataset. A domain gap was introduced by only using labels for sites located in the Western World, where geospatial data are typically less sparse than in the Global South. MMCR achieved an average F1 score of 0.444 when applied to sites located outside of the source domain, which is a considerable improvement to several supervised models (F1 scores between 0.107 and 0.424).

    The combined findings of this thesis contribute to the mapping and monitoring of cities on a global scale, which is crucial to support sustainable planning and urban SDG indicator monitoring.

  • Optically poled fibers for electro-optic applications

    Author: Joao Pereira
    Publication date: 2022-05-25 13:02

    The work presented in this thesis shows the development of optically poleddevices for use in electro-optic experiments. In the form of papers published,it describes three optical poling methods: green light poling (Paper I), poling with a UV lamp (Paper II), and corona discharge poling (Paper V).Applications using the poled components are studied in distributed sensingby exploring Rayleigh scattering in poled fibers (Paper III), intermodal interference in poled fibers (Paper IV), and FBG inscribed in poled fibers forvoltage sensing (Paper VI).In thermal poling, heat increases the mobility of added ions to the fiber.An external electric field displaces the charges creating a depletion regionclose to the anode, where the fiber core is usually positioned. The proximityof the metal electrode to the core can cause optical losses, making electrooptic applications less efficient. The need for additional dopant in the preformcan make the production of these devices expensive.Optical poling explores the presence of Ge E’ centers in the fiber core torelease charges after light excitation. These centers are already present in anyfiber with Ge in the core. This enables the development of a fiber for opticalpoling very similar to a standard telecom fiber, making it cheaper and easyto integrate with standard components. Optical poling does not rely on theformation of a depleted region in the cladding, and the core can be positionedfar from the metal electrodes. This advantage allows low-loss electro-opticcomponents to be fabricated.Optical poling is usually thought to have lower induced effects when compared with thermal poling. In this work, experiments with optical polingwere made to study the possibility of increasing the induced second-ordernonlinearities to a level comparable with thermal poling.The fabricated poled fibers were used to investigate their potential usein fiber sensing. The emphasis was to explore new technologies such as CPϕOTDR, few-mode fiber sensing, which gained attention in the latest years,and FBGs, which is a mature technology.The results presented in the Papers I-VI show the advances and potentialapplications explored.