Refine
H-BRS Bibliography
- yes (69) (remove)
Departments, institutes and facilities
- Fachbereich Informatik (45)
- Fachbereich Ingenieurwissenschaften und Kommunikation (14)
- Institut für Technik, Ressourcenschonung und Energieeffizienz (TREE) (10)
- Fachbereich Angewandte Naturwissenschaften (5)
- Fachbereich Wirtschaftswissenschaften (4)
- Institut für Cyber Security & Privacy (ICSP) (2)
- Fachbereich Sozialpolitik und Soziale Sicherung (1)
- Institut für funktionale Gen-Analytik (IFGA) (1)
- Institute of Visual Computing (IVC) (1)
Document Type
- Preprint (69) (remove)
Year of publication
Has Fulltext
- no (69) (remove)
Keywords
- Evolutionary Computation (2)
- inborn error of metabolism (2)
- ketone body (2)
- metabolic acidosis (2)
- metabolic decompensation (2)
- organic aciduria (2)
- unsupervised learning (2)
- ACAT1 (1)
- AML (1)
- Air Pollution Monitoring (1)
We introduce canonical weight normalization for convolutional neural networks. Inspired by the canonical tensor decomposition, we express the weight tensors in so-called canonical networks as scaled sums of outer vector products. In particular, we train network weights in the decomposed form, where scale weights are optimized separately for each mode. Additionally, similarly to weight normalization, we include a global scaling parameter. We study the initialization of the canonical form by running the power method and by drawing randomly from Gaussian or uniform distributions. Our results indicate that we can replace the power method with cheaper initializations drawn from standard distributions. The canonical re-parametrization leads to competitive normalization performance on the MNIST, CIFAR10, and SVHN data sets. Moreover, the formulation simplifies network compression. Once training has converged, the canonical form allows convenient model-compression by truncating the parameter sums.
Background There is a lack of cardiac magnetic resonance (CMR) data regarding mid- to long-term myocardial damage due to Covid-19 in elite athletes. Objective This study investigated mid-to long-term consequences of myocardial involvement after a Covid-19 infection in elite athletes.
Methods Between January 2020 and October 2021, 27 athletes of the German Olympic centre Rhineland with confirmed Covid-19 infection were analyzed. 9 healthy non-athlete volunteers served as control. CMR was performed in mean 182 days (SD 99) after initial positive test result.
Results CMR did not reveal any signs of acute myocarditis in regard to the current Lake Louise criteria or myocardial damage in any of the 26 elite athletes with previous Covid-19 infection. Nevertheless, 92 % of the athletes experienced a symptomatic course and 54 % reported lasting symptoms for more than 4 weeks. In one male athlete CMR revealed an arrhythmogenic right ventricular cardiomyopathy (ARVC) and this athlete was excluded from the study. Athletes had significantly enlarged left and right ventricle volumes and increased left ventricular myocardial mass in comparison to the healthy control group (LVEDVi 103.4 vs. 91.1 ml/m 2 p=0.031; RVEDVi 104.1 vs. 86.6 ml/m 2 p=0.007; and LVMi 59.0 vs. 46.2 g/m 2 p=0.002).
Conclusion Our findings suggest that the risk for mid-to long-term myocardial damage seems to be very low to negligible in elite athletes. No conclusions can be drawn regarding myocardial injury in the acute phase of infection nor about possible long-term myocardial effects in the general population.
TSEM: Temporally Weighted Spatiotemporal Explainable Neural Network for Multivariate Time Series
(2022)
Deep learning has become a one-size-fits-all solution for technical and business domains thanks to its flexibility and adaptability. It is implemented using opaque models, which unfortunately undermines the outcome trustworthiness. In order to have a better understanding of the behavior of a system, particularly one driven by time series, a look inside a deep learning model so-called posthoc eXplainable Artificial Intelligence (XAI) approaches, is important. There are two major types of XAI for time series data, namely model-agnostic and model-specific. Model-specific approach is considered in this work. While other approaches employ either Class Activation Mapping (CAM) or Attention Mechanism, we merge the two strategies into a single system, simply called the Temporally Weighted Spatiotemporal Explainable Neural Network for Multivariate Time Series (TSEM). TSEM combines the capabilities of RNN and CNN models in such a way that RNN hidden units are employed as attention weights for the CNN feature maps temporal axis. The result shows that TSEM outperforms XCM. It is similar to STAM in terms of accuracy, while also satisfying a number of interpretability criteria, including causality, fidelity, and spatiotemporality.
Self-supervised learning has proved to be a powerful approach to learn image representations without the need of large labeled datasets. For underwater robotics, it is of great interest to design computer vision algorithms to improve perception capabilities such as sonar image classification. Due to the confidential nature of sonar imaging and the difficulty to interpret sonar images, it is challenging to create public large labeled sonar datasets to train supervised learning algorithms. In this work, we investigate the potential of three self-supervised learning methods (RotNet, Denoising Autoencoders, and Jigsaw) to learn high-quality sonar image representation without the need of human labels. We present pre-training and transfer learning results on real-life sonar image datasets. Our results indicate that self-supervised pre-training yields classification performance comparable to supervised pre-training in a few-shot transfer learning setup across all three methods. Code and self-supervised pre-trained models are be available at https://github.com/agrija9/ssl-sonar-images
Comparative study of 3D object detection frameworks based on LiDAR data and sensor fusion techniques
(2022)
Estimating and understanding the surroundings of the vehicle precisely forms the basic and crucial step for the autonomous vehicle. The perception system plays a significant role in providing an accurate interpretation of a vehicle's environment in real-time. Generally, the perception system involves various subsystems such as localization, obstacle (static and dynamic) detection, and avoidance, mapping systems, and others. For perceiving the environment, these vehicles will be equipped with various exteroceptive (both passive and active) sensors in particular cameras, Radars, LiDARs, and others. These systems are equipped with deep learning techniques that transform the huge amount of data from the sensors into semantic information on which the object detection and localization tasks are performed. For numerous driving tasks, to provide accurate results, the location and depth information of a particular object is necessary. 3D object detection methods, by utilizing the additional pose data from the sensors such as LiDARs, stereo cameras, provides information on the size and location of the object. Based on recent research, 3D object detection frameworks performing object detection and localization on LiDAR data and sensor fusion techniques show significant improvement in their performance. In this work, a comparative study of the effect of using LiDAR data for object detection frameworks and the performance improvement seen by using sensor fusion techniques are performed. Along with discussing various state-of-the-art methods in both the cases, performing experimental analysis, and providing future research directions.
Ice accumulation in the blades of wind turbines can cause them to describe anomalous rotations or no rotations at all, thus affecting the generation of electricity and power output. In this work, we investigate the problem of ice accumulation in wind turbines by framing it as anomaly detection of multi-variate time series. Our approach focuses on two main parts: first, learning low-dimensional representations of time series using a Variational Recurrent Autoencoder (VRAE), and second, using unsupervised clustering algorithms to classify the learned representations as normal (no ice accumulated) or abnormal (ice accumulated). We have evaluated our approach on a custom wind turbine time series dataset, for the two-classes problem (one normal versus one abnormal class), we obtained a classification accuracy of up to 96$\%$ on test data. For the multiple-class problem (one normal versus multiple abnormal classes), we present a qualitative analysis of the low-dimensional learned latent space, providing insights into the capacities of our approach to tackle such problem. The code to reproduce this work can be found here https://github.com/agrija9/Wind-Turbines-VRAE-Paper.
It has been well proved that deep networks are efficient at extracting features from a given (source) labeled dataset. However, it is not always the case that they can generalize well to other (target) datasets which very often have a different underlying distribution. In this report, we evaluate four different domain adaptation techniques for image classification tasks: DeepCORAL, DeepDomainConfusion, CDAN and CDAN+E. These techniques are unsupervised given that the target dataset dopes not carry any labels during training phase. We evaluate model performance on the office-31 dataset. A link to the github repository of this report can be found here: https://github.com/agrija9/Deep-Unsupervised-Domain-Adaptation.
Recent experimental evidence suggest that mebendazole, a popular antiparasitic drug, binds to heat shock protein 90 (Hsp90) and inhibit acute myeloid leukemia cell growth. In this study we use quantum mechanics (QM), molecular similarity and molecular dynamics (MD) calculations to predict possible binding poses of mebendazole to the adenosine triphosphate (ATP) binding site of Hsp90. Extensive conformational searches and minimization of the five tautomers of mebendazole using MP2/aug-cc-pVTZ theory level resulting in 152 minima being identified. Mebendazole-Hsp90 complex models were created using the QM optimized conformations and protein coordinates obtained from experimental crystal structures that were chosen through similarity calculations. Nine different poses were identified from a total of 600 ns of explicit solvent, all-atom MD simulations using two different force fields. All simulations support the hypothesis that mebendazole is able to bind to the ATP binding site of Hsp90.
Urban LoRa networks promise to provide a cost-efficient and scalable communication backbone for smart cities. One core challenge in rolling out and operating these networks is radio network planning, i.e., precise predictions about possible new locations and their impact on network coverage. Path loss models aid in this task, but evaluating and comparing different models requires a sufficiently large set of high-quality received packet power samples. In this paper, we report on a corresponding large-scale measurement study covering an urban area of 200km2 over a period of 230 days using sensors deployed on garbage trucks, resulting in more than 112 thousand high-quality samples for received packet power. Using this data, we compare eleven previously proposed path loss models and additionally provide new coefficients for the Log-distance model. Our results reveal that the Log-distance model and other well-known empirical models such as Okumura or Winner+ provide reasonable estimations in an urban environment, and terrain based models such as ITM or ITWOM have no advantages. In addition, we derive estimations for the needed sample size in similar measurement campaigns. To stimulate further research in this direction, we make all our data publicly available.
The majority of biomedical knowledge is stored in structured databases or as unstructured text in scientific publications. This vast amount of information has led to numerous machine learning-based biological applications using either text through natural language processing (NLP) or structured data through knowledge graph embedding models (KGEMs). However, representations based on a single modality are inherently limited. To generate better representations of biological knowledge, we propose STonKGs, a Sophisticated Transformer trained on biomedical text and Knowledge Graphs. This multimodal Transformer uses combined input sequences of structured information from KGs and unstructured text data from biomedical literature to learn joint representations. First, we pre-trained STonKGs on a knowledge base assembled by the Integrated Network and Dynamical Reasoning Assembler (INDRA) consisting of millions of text-triple pairs extracted from biomedical literature by multiple NLP systems. Then, we benchmarked STonKGs against two baseline models trained on either one of the modalities (i.e., text or KG) across eight different classification tasks, each corresponding to a different biological application. Our results demonstrate that STonKGs outperforms both baselines, especially on the more challenging tasks with respect to the number of classes, improving upon the F1-score of the best baseline by up to 0.083. Additionally, our pre-trained model as well as the model architecture can be adapted to various other transfer learning applications. Finally, the source code and pre-trained STonKGs models are available at https://github.com/stonkgs/stonkgs and https://huggingface.co/stonkgs/stonkgs-150k.
Application of underwater robots are on the rise, most of them are dependent on sonar for underwater vision, but the lack of strong perception capabilities limits them in this task. An important issue in sonar perception is matching image patches, which can enable other techniques like localization, change detection, and mapping. There is a rich literature for this problem in color images, but for acoustic images, it is lacking, due to the physics that produce these images. In this paper we improve on our previous results for this problem (Valdenegro-Toro et al, 2017), instead of modeling features manually, a Convolutional Neural Network (CNN) learns a similarity function and predicts if two input sonar images are similar or not. With the objective of improving the sonar image matching problem further, three state of the art CNN architectures are evaluated on the Marine Debris dataset, namely DenseNet, and VGG, with a siamese or two-channel architecture, and contrastive loss. To ensure a fair evaluation of each network, thorough hyper-parameter optimization is executed. We find that the best performing models are DenseNet Two-Channel network with 0.955 AUC, VGG-Siamese with contrastive loss at 0.949 AUC and DenseNet Siamese with 0.921 AUC. By ensembling the top performing DenseNet two-channel and DenseNet-Siamese models overall highest prediction accuracy obtained is 0.978 AUC, showing a large improvement over the 0.91 AUC in the state of the art.
The ability to discriminate between different ionic species, termed ion selectivity, is a key feature of ion channels and forms the basis for their physiological function. Members of the degenerin/epithelial sodium channel (DEG/ENaC) superfamily of trimeric ion channels are typically sodium selective, but to a surprisingly variable degree. While acid-sensing ion channels (ASICs) are weakly sodium selective (sodium:potassium around 10:1), ENaCs show a remarkably high preference for sodium over potassium (>500:1). The most obvious explanation for this discrepancy may be expected to originate from differences in the pore-lining second transmembrane segment (M2). However, these show a relatively high degree of sequence conservation between ASICs and ENaCs and previous functional and structural studies could not unequivocally establish that differences in M2 alone can account for the disparate degrees of ion selectivity. By contrast, surprisingly little is known about the contributions of the first transmembrane segment (M1) and the preceding pre-M1 region. In this study, we use conventional and non-canonical amino acid-based mutagenesis in combination with a variety of electrophysiological approaches to show that the pre-M1 and M1 regions of mASIC1a channels are major determinants of ion selectivity. Mutational investigations of the corresponding regions in hENaC show that they contribute less to ion selectivity, despite affecting ion conductance. In conclusion, our work supports the notion that the remarkably different degrees of sodium selectivity in ASICs and ENaCs are achieved through different mechanisms. The results further highlight how M1 and pre-M1 are likely to differentially affect pore structure in these related channels.
The lattice Boltzmann method (LBM) is an efficient simulation technique for computational fluid mechanics and beyond. It is based on a simple stream-and-collide algorithm on Cartesian grids, which is easily compatible with modern machine learning architectures. While it is becoming increasingly clear that deep learning can provide a decisive stimulus for classical simulation techniques, recent studies have not addressed possible connections between machine learning and LBM. Here, we introduce Lettuce, a PyTorch-based LBM code with a threefold aim. Lettuce enables GPU accelerated calculations with minimal source code, facilitates rapid prototyping of LBM models, and enables integrating LBM simulations with PyTorch's deep learning and automatic differentiation facility. As a proof of concept for combining machine learning with the LBM, a neural collision model is developed, trained on a doubly periodic shear layer and then transferred to a different flow, a decaying turbulence. We also exemplify the added benefit of PyTorch's automatic differentiation framework in flow control and optimization. To this end, the spectrum of a forced isotropic turbulence is maintained without further constraining the velocity field.
The Covid-19 pandemic has challenged educators across the world to move their teaching and mentoring from in-person to remote. During nonpandemic semesters at their institutes (e.g. universities), educators can directly provide students the software environment needed to support their learning - either in specialized computer laboratories (e.g. computational chemistry labs) or shared computer spaces. These labs are often supported by staff that maintains the operating systems (OS) and software. But how does one provide a specialized software environment for remote teaching? One solution is to provide students a customized operating system (e.g., Linux) that includes open-source software for supporting your teaching goals. However, such a solution should not require students to install the OS alongside their existing one (i.e. dual/multi-booting) or be used as a complete replacement. Such approaches are risky because of a) the students' possible lack of software expertise, b) the possible disruption of an existing software workflow that is needed in other classes or by other family members, and c) the importance of maintaining a working computer when isolated (e.g. societal restrictions). To illustrate possible solutions, we discuss our approach that used a customized Linux OS and a Docker container in a course that teaches computational chemistry and Python3.
Off-lattice Boltzmann methods increase the flexibility and applicability of lattice Boltzmann methods by decoupling the discretizations of time, space, and particle velocities. However, the velocity sets that are mostly used in off-lattice Boltzmann simulations were originally tailored to on-lattice Boltzmann methods. In this contribution, we show how the accuracy and efficiency of weakly and fully compressible semi-Lagrangian off-lattice Boltzmann simulations is increased by velocity sets derived from cubature rules, i.e. multivariate quadratures, which have not been produced by the Gauss-product rule. In particular, simulations of 2D shock-vortex interactions indicate that the cubature-derived degree-nine D2Q19 velocity set is capable to replace the Gauss-product rule-derived D2Q25. Likewise, the degree-five velocity sets D3Q13 and D3Q21, as well as a degree-seven D3V27 velocity set were successfully tested for 3D Taylor-Green vortex flows to challenge and surpass the quality of the customary D3Q27 velocity set. In compressible 3D Taylor-Green vortex flows with Mach numbers Ma={0.5;1.0;1.5;2.0} on-lattice simulations with velocity sets D3Q103 and D3V107 showed only limited stability, while the off-lattice degree-nine D3Q45 velocity set accurately reproduced the kinetic energy provided by literature.
Describing the elephant: a foundational model of human needs, motivation, behaviour, and wellbeing
(2020)
Models of basic psychological needs have been present and popular in the academic and lay literature for more than a century yet reviews of needs models show an astonishing lack of consensus. This raises the question of what basic human psychological needs are and if this can be consolidated into a model or framework that can align previous research and empirical study. The authors argue that the lack of consensus arises from researchers describing parts of the proverbial elephant correctly but failing to describe the full elephant. Through redefining what human needs are and matching this to an evolutionary framework we can see broad consensus across needs models and neatly slot constructs and psychological and behavioural theories into this framework. This enables a descriptive model of drives, motives, and well-being that can be simply outlined but refined enough to do justice to the complexities of human behaviour. This also raises some issues of how subjective well-being is and should be measured. Further avenues of research and how to continue building this model and framework are proposed.
Fundamental hydrogen storage properties of TiFe-alloy with partial substitution of Fe by Ti and Mn
(2020)
TiFe intermetallic compound has been extensively studied, owing to its low cost, good volumetric hydrogen density, and easy tailoring of hydrogenation thermodynamics by elemental substitution. All these positive aspects make this material promising for large-scale applications of solid-state hydrogen storage. On the other hand, activation and kinetic issues should be amended and the role of elemental substitution should be further understood. This work investigates the thermodynamic changes induced by the variation of Ti content along the homogeneity range of the TiFe phase (Ti:Fe ratio from 1:1 to 1:0.9) and of the substitution of Mn for Fe between 0 and 5 at.%. In all considered alloys, the major phase is TiFe-type together with minor amounts of TiFe2 or \b{eta}-Ti-type and Ti4Fe2O-type at the Ti-poor and rich side of the TiFe phase domain, respectively. Thermodynamic data agree with the available literature but offer here a comprehensive picture of hydrogenation properties over an extended Ti and Mn compositional range. Moreover, it is demonstrated that Ti-rich alloys display enhanced storage capacities, as long as a limited amount of \b{eta}-Ti is formed. Both Mn and Ti substitutions increase the cell parameter by possibly substituting Fe, lowering the plateau pressures and decreasing the hysteresis of the isotherms. A full picture of the dependence of hydrogen storage properties as a function of the composition will be discussed, together with some observed correlations.
Object detectors have improved considerably in the last years by using advanced CNN architectures. However, many detector hyper-parameters are generally manually tuned, or they are used with values set by the detector authors. Automatic Hyper-parameter optimization has not been explored in improving CNN-based object detectors hyper-parameters. In this work, we propose the use of Black-box optimization methods to tune the prior/default box scales in Faster R-CNN and SSD, using Bayesian Optimization, SMAC, and CMA-ES. We show that by tuning the input image size and prior box anchor scale on Faster R-CNN mAP increases by 2% on PASCAL VOC 2007, and by 3% with SSD. On the COCO dataset with SSD there are mAP improvement in the medium and large objects, but mAP decreases by 1% in small objects. We also perform a regression analysis to find the significant hyper-parameters to tune.
In this paper we introduce the Perception for Autonomous Systems (PAZ) software library. PAZ is a hierarchical perception library that allow users to manipulate multiple levels of abstraction in accordance to their requirements or skill level. More specifically, PAZ is divided into three hierarchical levels which we refer to as pipelines, processors, and backends. These abstractions allows users to compose functions in a hierarchical modular scheme that can be applied for preprocessing, data-augmentation, prediction and postprocessing of inputs and outputs of machine learning (ML) models. PAZ uses these abstractions to build reusable training and prediction pipelines for multiple robot perception tasks such as: 2D keypoint estimation, 2D object detection, 3D keypoint discovery, 6D pose estimation, emotion classification, face recognition, instance segmentation, and attention mechanisms.