Filters








4,418 Hits in 3.1 sec

Wasserstein-based Graph Alignment [article]

Hermina Petric Maretic, Mireille El Gheche, Matthias Minder, Giovanni Chierchia, Pascal Frossard
2020 arXiv   pre-print
We propose a novel method for comparing non-aligned graphs of different sizes, based on the Wasserstein distance between graph signal distributions induced by the respective graph Laplacian matrices. Specifically, we cast a new formulation for the one-to-many graph alignment problem, which aims at matching a node in the smaller graph with one or more nodes in the larger graph. By integrating optimal transport in our graph comparison framework, we generate both a structurally-meaningful graph
more » ... tance, and a signal transportation plan that models the structure of graph data. The resulting alignment problem is solved with stochastic gradient descent, where we use a novel Dykstra operator to ensure that the solution is a one-to-many (soft) assignment matrix. We demonstrate the performance of our novel framework on graph alignment and graph classification, and we show that our method leads to significant improvements with respect to the state-of-the-art algorithms for each of these tasks.
arXiv:2003.06048v1 fatcat:ytbhhkccnbdfziccopfruh7ygm

Virtual reality explored

Matthias Minderer, Christopher D. Harvey, Flavio Donato, Edvard I. Moser
2016 Nature  
Matthias Minderer and • • In most rodent set-ups, the animal receives visual information from an immersive screen that spans its field of vision.  ... 
doi:10.1038/nature17899 pmid:27193673 fatcat:s2rishu7wrajffidazxfain3n4

SCENIC: A JAX Library for Computer Vision Research and Beyond [article]

Mostafa Dehghani and Alexey Gritsenko and Anurag Arnab and Matthias Minderer and Yi Tay
2021 arXiv   pre-print
Scenic is an open-source JAX library with a focus on Transformer-based models for computer vision research and beyond. The goal of this toolkit is to facilitate rapid experimentation, prototyping, and research of new vision architectures and models. Scenic supports a diverse range of vision tasks (e.g., classification, segmentation, detection)and facilitates working on multi-modal problems, along with GPU/TPU support for multi-host, multi-device large-scale training. Scenic also offers
more » ... implementations of state-of-the-art research models spanning a wide range of modalities. Scenic has been successfully used for numerous projects and published papers and continues serving as the library of choice for quick prototyping and publication of new research ideas.
arXiv:2110.11403v1 fatcat:x2t6wy54vzh3zmuj5poh6qehkm

Overcoming cut-off restrictions in multimorbidity prevalence estimates

Barbara M Holzer, Klarissa Siebenhuener, Matthias Bopp, Christoph E Minder
2014 BMC Public Health  
BACKGROUND: Presently, there is no consensus on how to define multimorbidity. In this paper we investigate the connection between prevalence estimates for two or more and three or more chronic conditions to improve comparability of multimorbidity studies with different cut-offs. METHODS: In a systematic review of the literature published between January, 1990 and December, 2011, we found 52 suitable studies, many providing prevalence estimates for several age groups. A total of 31 studies
more » ... ed both the prevalence for multimorbidity based on two or more chronic conditions and three or more chronic conditions, which were analysed in this study. Our research question was whether there is a systematic interrelation between these two prevalence estimates, and how this could be used to improve the comparability of studies on the burden of multimorbidity. RESULTS: Actually, we found a tight relationship between the prevalence of two or more and three or more chronic conditions. Moreover, each of these estimates can be predicted from the other. I.e. the cut-offs of two or three for the number of chronic conditions produce essentially the same information on prevalence. CONCLUSIONS: Our study shows a way to enhance and improve the comparability of prevalence estimates from different multimorbidity studies.
doi:10.1186/1471-2458-14-780 pmid:25084825 pmcid:PMC4133617 fatcat:bnxmqn2h2zh35hj5k5msrrelka

Generic Routing Metric and Policies for WSNs [chapter]

Olga Saukh, Pedro José Marrón, Andreas Lachenmann, Matthias Gauger, Daniel Minder, Kurt Rothermel
2006 Lecture Notes in Computer Science  
Energy-aware algorithms have proven to be a crucial part of sensor network applications, especially if they are required to operate for extended periods of time. Among these, efficient routing algorithms are of utter importance since their effect can be experienced by all other layers. Thus, the optimization and accurate prediction of the lifetime of the system can only be performed in the presence of accurate execution models that take energy consumption into account. In this paper, we propose
more » ... a generic routing metric and associated policies that encompass most other existing metrics in the literature and use this model for the optimal construction of a routing tree to the sink. We also provide experimental results that show the benefits of using our novel metric.
doi:10.1007/11669463_10 fatcat:rpxdixonuvg4njmo6pg2575yma

Characterization of a Parabolic trough Collector for Process Heat Applications

Marco Larcher, Matthias Rommel, Andreas Bohren, Elimar Frank, Stefan Minder
2014 Energy Procedia  
In this paper we present experimental investigations on a parabolic trough collector for process heat applications under development which has an aperture width of about 1.845 m. The collector which is installed at the SPF outdoor test laboratory has a length of 10 m. As a final goal, our investigations aim at the possibility to elaborate a valid efficiency curve of the collector from measurements of the optical behavior on the one hand (η 0 -measurements, efficiency measurements carried out at
more » ... low and high operating temperatures) and on the other hand heat loss measurements of the receiver, carried out at high collector operating temperatures. This paper reports on preliminary results of collector efficiency measurements carried out for collector operating temperatures up to about 200°C and on measurements to determine the incidence angle modifier curve (IAM) using steady state measurements conditions. This paper is an addition to a prior publication [2] . There the collector size was smaller and our new test rig was not yet available.
doi:10.1016/j.egypro.2014.10.313 fatcat:4wakr323eja2vnouq2vdahjfam

Passive surveillance of Leptospira infection in swine in Germany

Katrin Strutzberg-Minder, Astrid Tschentscher, Martin Beyerbach, Matthias Homuth, Lothar Kreienbrock
2018 Porcine Health Management  
As no current data are available on the prevalence of leptospiral infection in swine in Germany, we analysed laboratory data from diagnostic examinations carried out on samples from swine all over Germany from January 2011 to September 2016. A total of 29,829 swine sera were tested by microscopic agglutination test (MAT) for antibodies against strains of eleven Leptospira serovars. Results: Overall, 20.2% (6025) of the total sample collection tested positive for leptospiral infection.
more » ... vity ranged between 16.3% (964) in 2011 and 30.9% (941) in 2016 (January to September only). Of all samples, 11.6% (57.3% of the positives) reacted with only one Leptospira serovar, and only 8.6% (42.7% of the positives) reacted simultaneously with two or more serovars. The most frequently detected serovar was Bratislava, which was found in 11.6% (3448) of all samples, followed by the serovars Australis in 7.. Modelling shows that both the year and the reason for testing at the laboratory had statistically strong effects on the test results; however, no interactions were determined between those factors. The results support the suggestion that the seropositivities found may be considered to indicate the state of leptospiral infections in the German swine population. Conclusion: Although data from passive surveillance are prone to selection bias, stratified analysis by initial reason for examination and analyses by model approaches may correct for biases. A prevalence of about 20% for a leptospiral infection is most probable for sows with reproductive problems in Germany, with an increasing trend. Swine in Germany are probably a reservoir host for serovar Bratislava, but in contrast to other studies not for Pomona and Tarassovi.
doi:10.1186/s40813-018-0086-5 pmid:29610674 pmcid:PMC5872537 fatcat:qsrjfkepurfs5ahagfcucialha

Educational inequalities in life expectancy in German speaking part of Switzerland 1990-1997: Swiss National Cohort

Adrian Spoerri, Marcel Zwahlen, Matthias Egger, Felix Gutzwiller, Christoph Minder, Matthias Bopp
2006 Swiss Medical Weekly  
Matthias Bopp Department of Social & Preventive Medicine Hirschengraben 84 CH-8006 Zürich Switzerland E-Mail: bopp@ifspm.unizh.ch  ... 
pmid:16633959 fatcat:e2k2ouaqb5cnvgcwtypu6su4f4

Dynamic Reorganization of Neuronal Activity Patterns in Parietal Cortex

Laura N. Driscoll, Noah L. Pettit, Matthias Minderer, Selmaan N. Chettih, Christopher D. Harvey
2017 Cell  
Neuronal representations change as associations are learned between sensory stimuli and behavioral actions. However, it is poorly understood whether representations for learned associations stabilize in cortical association areas or continue to change following learning. We tracked the activity of posterior parietal cortex neurons for a month as mice stably performed a virtual-navigation task. The relationship between cells' activity and task features was mostly stable on single days but
more » ... nt major reorganization over weeks. The neurons informative about task features (trial type and maze locations) changed across days. Despite changes in individual cells, the population activity had statistically similar properties each day and stable information for over a week. As mice learned additional associations, new activity patterns emerged in the neurons used for existing representations without greatly affecting the rate of change of these representations. We propose that dynamic neuronal activity patterns could balance plasticity for learning and stability for memory. eTOC 2 Contrary to the idea that representations of sensory stimuli or the activity patterns that accompany motor actions are stable, neuronal representations in the parietal cortex can change across days, possibly allowing for the tradeoff between stable encoding of information and flexibility for incorporating new information. Driscoll et al.
doi:10.1016/j.cell.2017.07.021 pmid:28823559 pmcid:PMC5718200 fatcat:4lex74serzhr7gsrti7mq7dkaa

Revisiting the Calibration of Modern Neural Networks [article]

Matthias Minderer, Josip Djolonga, Rob Romijnders, Frances Hubis, Xiaohua Zhai, Neil Houlsby, Dustin Tran, Mario Lucic
2021 arXiv   pre-print
Accurate estimation of predictive uncertainty (model calibration) is essential for the safe application of neural networks. Many instances of miscalibration in modern neural networks have been reported, suggesting a trend that newer, more accurate models produce poorly calibrated predictions. Here, we revisit this question for recent state-of-the-art image classification models. We systematically relate model calibration and accuracy, and find that the most recent models, notably those not
more » ... convolutions, are among the best calibrated. Trends observed in prior model generations, such as decay of calibration with distribution shift or model size, are less pronounced in recent architectures. We also show that model size and amount of pretraining do not fully explain these differences, suggesting that architecture is a major determinant of calibration properties.
arXiv:2106.07998v2 fatcat:rod3flaqujcc3ln6ve4o4z5zsm

The Spatial Structure of Neural Encoding in Mouse Posterior Cortex during Navigation

Matthias Minderer, Kristen D. Brown, Christopher D. Harvey
2019 Neuron  
., 2012; Minderer et al., 2016) . The same locomotor actions resulted in distinct optic flow patterns and vice versa.  ... 
doi:10.1016/j.neuron.2019.01.029 pmid:30772081 pmcid:PMC6642748 fatcat:rjzunm2kdjfrbhfk7pgxk3aoru

Management and configuration issues for sensor networks

Pedro José Marrón, Andreas Lachenmann, Daniel Minder, Matthias Gauger, Olga Saukh, Kurt Rothermel
2005 International Journal of Network Management  
Short abstract We define three of the key issues related to efficient management and configuration of sensor networks: the distribution and management of roles within the network, efficient code distribution algorithms, and efficient on-the-fly code update algorithms. We present some results for each of these issues as we have dealt with them within the TinyCubus project. Full abstract In this paper, we define three of the key issues that need to be solved in order to provide efficient
more » ... t and configuration of applications and system software in sensor networks: the distribution and management of roles within the network, efficient code distribution algorithms, and efficient on-the-fly code update algorithms for sensor networks. The first issue is motivated by the increasing heterogeneity of sensor network applications and their need for more complex (non-homogeneous) network topologies and structures. The second one is motivated by the intrinsic energy constraint issues and, in general, the resource limitation of sensor networks. Finally, the third one is needed due to the nature of monitoring applications and optimization needs from applications that should be able to efficiently incorporate code updates so that the network can adapt to its surroundings on the fly. In this paper we present related work and some results for each of these issues as we have dealt with them within the TinyCubus project.
doi:10.1002/nem.571 fatcat:sp45ltexqjcfhi7g4frn67jlpa

Simple Open-Vocabulary Object Detection with Vision Transformers [article]

Matthias Minderer, Alexey Gritsenko, Austin Stone, Maxim Neumann, Dirk Weissenborn, Alexey Dosovitskiy, Aravindh Mahendran, Anurag Arnab, Mostafa Dehghani, Zhuoran Shen, Xiao Wang, Xiaohua Zhai (+2 others)
2022 arXiv   pre-print
Combining simple architectures with large-scale pre-training has led to massive improvements in image classification. For object detection, pre-training and scaling approaches are less well established, especially in the long-tailed and open-vocabulary setting, where training data is relatively scarce. In this paper, we propose a strong recipe for transferring image-text models to open-vocabulary object detection. We use a standard Vision Transformer architecture with minimal modifications,
more » ... rastive image-text pre-training, and end-to-end detection fine-tuning. Our analysis of the scaling properties of this setup shows that increasing image-level pre-training and model size yield consistent improvements on the downstream detection task. We provide the adaptation strategies and regularizations needed to attain very strong performance on zero-shot text-conditioned and one-shot image-conditioned object detection. Code and models are available on GitHub.
arXiv:2205.06230v1 fatcat:q6zlylwljfadfoxii5dl25puiu

Automatic Shortcut Removal for Self-Supervised Representation Learning [article]

Matthias Minderer, Olivier Bachem, Neil Houlsby, Michael Tschannen
2020 arXiv   pre-print
Correspondence to: Matthias Minderer <mjlm@google.com>. We address the unwanted removal of potentially useful features in Section 3.1.  ... 
arXiv:2002.08822v3 fatcat:xxzqyaocjrbmnlp5vydpjigbmm

Evidence-based design recommendations for prevalence studies on multimorbidity: improving comparability of estimates

Barbara M. Holzer, Klarissa Siebenhuener, Matthias Bopp, Christoph E. Minder
2017 Population Health Metrics  
In aging populations, multimorbidity causes a disease burden of growing importance and cost. However, estimates of the prevalence of multimorbidity (prevMM) vary widely across studies, impeding valid comparisons and interpretation of differences. With this study we pursued two research objectives: (1) to identify a set of study design and demographic factors related to prevMM, and (2) based on (1), to formulate design recommendations for future studies with improved comparability of prevalence
more » ... stimates. Methods: Study data were obtained through systematic review of the literature. UsingPubMed/MEDLINE, Embase, CINAHL, Web of Science, BIOSIS, and Google Scholar, we looked for articles with the terms "multimorbidity," "comorbidity," "polymorbidity," and variations of these published in English or German in the years 1990 to 2011. We selected quantitative studies of the prevalence of multimorbidity (two or more chronic medical conditions) with a minimum sample size of 50 and a study population with a majority of Caucasians. Our database consisted of prevalence estimates in 108 age groups taken from 45 studies. To assess the effects of study design variables, we used meta regression models. Results: In 58% of the studies, there was only one age group, i.e., no stratification by age. The number of persons per age group ranged from 136 to 5.6 million. Our analyses identified the following variables as highly significant: "mean age," "number of age groups", and "data reporting quality" (all p < 0.0001). "Setting," "disease classification," and "number of diseases in the classification" were significant (0.01 < p ≤ 0.03), and "data collection period" and "data source" were non-significant. A separate analysis showed that prevMM was significantly higher in women than men (sign test, p = 0.0015). Conclusions: Comparable prevalence estimates are urgently needed for realistic description of the magnitude of the problem of multimorbidity. Based on the results of our analyses of variables affecting prevMM, we make some design recommendations. Our suggestions were guided by a pragmatic approach and aimed at facilitating the implementation of a uniform methodology. This should aid progress towards a more uniform operationalization of multimorbidity.
doi:10.1186/s12963-017-0126-4 pmid:28270157 pmcid:PMC5341353 fatcat:izjik5pdjfcydkwmbylf2ls4xi
« Previous Showing results 1 — 15 out of 4,418 results