error modelling Dotaz Zobrazit nápovědu
Supervised learning of perceptron networks is investigated as an optimization problem. It is shown that both the theoretical and the empirical error functionals achieve minima over sets of functions computable by networks with a given number n of perceptrons. Upper bounds on rates of convergence of these minima with n increasing are derived. The bounds depend on a certain regularity of training data expressed in terms of variational norms of functions interpolating the data (in the case of the empirical error) and the regression function (in the case of the expected error). Dependence of this type of regularity on dimensionality and on magnitudes of partial derivatives is investigated. Conditions on the data, which guarantee that a good approximation of global minima of error functionals can be achieved using networks with a limited complexity, are derived. The conditions are in terms of oscillatory behavior of the data measured by the product of a function of the number of variables d, which is decreasing exponentially fast, and the maximum of the magnitudes of the squares of the L(1)-norms of the iterated partial derivatives of the order d of the regression function or some function, which interpolates the sample of the data. The results are illustrated by examples of data with small and high regularity constructed using Boolean functions and the gaussian function.
Rainfall data are a crucial input for various tasks concerning the wet weather period. Nevertheless, their measurement is affected by random and systematic errors that cause an underestimation of the rainfall volume. Therefore, the general objective of the presented work was to assess the credibility of measured rainfall data and to evaluate the effect of measurement errors on urban drainage modelling tasks. Within the project, the methodology of the tipping bucket rain gauge (TBR) was defined and assessed in terms of uncertainty analysis. A set of 18 TBRs was calibrated and the results were compared to the previous calibration. This enables us to evaluate the ageing of TBRs. A propagation of calibration and other systematic errors through the rainfall-runoff model was performed on experimental catchment. It was found that the TBR calibration is important mainly for tasks connected with the assessment of peak values and high flow durations. The omission of calibration leads to up to 30% underestimation and the effect of other systematic errors can add a further 15%. The TBR calibration should be done every two years in order to catch up the ageing of TBR mechanics. Further, the authors recommend to adjust the dynamic test duration proportionally to generated rainfall intensity.
We investigate footprint geolocation uncertainties of a spectroradiometer mounted on an unmanned aircraft system (UAS). Two microelectromechanical systems-based inertial measurement units (IMUs) and global navigation satellite system (GNSS) receivers were used to determine the footprint location and extent of the spectroradiometer. Errors originating from the on-board GNSS/IMU sensors were propagated through an aerial data georeferencing model, taking into account a range of values for the spectroradiometer field of view (FOV), integration time, UAS flight speed, above ground level (AGL) flying height, and IMU grade. The spectroradiometer under nominal operating conditions (8 ∘ FOV, 10 m AGL height, 0.6 s integration time, and 3 m/s flying speed) resulted in footprint extent of 140 cm across-track and 320 cm along-track, and a geolocation uncertainty of 11 cm. Flying height and orientation measurement accuracy had the largest influence on the geolocation uncertainty, whereas the FOV, integration time, and flying speed had the biggest impact on the size of the footprint. Furthermore, with an increase in flying height, the rate of increase in geolocation uncertainty was found highest for a low-grade IMU. To increase the footprint geolocation accuracy, we recommend reducing flying height while increasing the FOV which compensates the footprint area loss and increases the signal strength. The disadvantage of a lower flying height and a larger FOV is a higher sensitivity of the footprint size to changing distance from the target. To assist in matching the footprint size to uncertainty ratio with an appropriate spatial scale, we list the expected ratio for a range of IMU grades, FOVs and AGL heights.
- Klíčová slova
- UAS, UAV, aerial spectroscopy, error propagation, footprint, geolocation, spectroradiometer,
- Publikační typ
- časopisecké články MeSH
Path integration is thought to rely on vestibular and proprioceptive cues yet most studies in humans involve primarily visual input, providing limited insight into their respective contributions. We developed a paradigm involving walking in an omnidirectional treadmill in which participants were guided on two sides of a triangle and then found their back way to origin. In Experiment 1, we tested a range of different triangle types while keeping the distance of the unguided side constant to determine the influence of spatial geometry. Participants overshot the angle they needed to turn and undershot the distance they needed to walk, with no consistent effect of triangle type. In Experiment 2, we manipulated distance while keeping angle constant to determine how path integration operated over both shorter and longer distances. Participants underestimated the distance they needed to walk to the origin, with error increasing as a function of the walked distance. To attempt to account for our findings, we developed configural-based computational models involving vector addition, the second of which included terms for the influence of past trials on the current one. We compared against a previously developed configural model of human path integration, the Encoding-Error model. We found that the vector addition models captured the tendency of participants to under-encode guided sides of the triangles and an influence of past trials on current trials. Together, our findings expand our understanding of body-based contributions to human path integration, further suggesting the value of vector addition models in understanding these important components of human navigation.
- MeSH
- chůze fyziologie MeSH
- dospělí MeSH
- lidé MeSH
- orientace fyziologie MeSH
- podněty MeSH
- propriocepce fyziologie MeSH
- prostorová navigace fyziologie MeSH
- teoretické modely MeSH
- vnímání prostoru fyziologie MeSH
- výpočetní biologie metody MeSH
- Check Tag
- dospělí MeSH
- lidé MeSH
- mužské pohlaví MeSH
- ženské pohlaví MeSH
- Publikační typ
- časopisecké články MeSH
- Research Support, U.S. Gov't, Non-P.H.S. MeSH
To gain a better understanding of the global application of soil erosion prediction models, we comprehensively reviewed relevant peer-reviewed research literature on soil-erosion modelling published between 1994 and 2017. We aimed to identify (i) the processes and models most frequently addressed in the literature, (ii) the regions within which models are primarily applied, (iii) the regions which remain unaddressed and why, and (iv) how frequently studies are conducted to validate/evaluate model outcomes relative to measured data. To perform this task, we combined the collective knowledge of 67 soil-erosion scientists from 25 countries. The resulting database, named 'Global Applications of Soil Erosion Modelling Tracker (GASEMT)', includes 3030 individual modelling records from 126 countries, encompassing all continents (except Antarctica). Out of the 8471 articles identified as potentially relevant, we reviewed 1697 appropriate articles and systematically evaluated and transferred 42 relevant attributes into the database. This GASEMT database provides comprehensive insights into the state-of-the-art of soil- erosion models and model applications worldwide. This database intends to support the upcoming country-based United Nations global soil-erosion assessment in addition to helping to inform soil erosion research priorities by building a foundation for future targeted, in-depth analyses. GASEMT is an open-source database available to the entire user-community to develop research, rectify errors, and make future expansions.
- Klíčová slova
- Erosion rates, GIS, Land degradation, Land sustainability, Modelling, Policy support,
- Publikační typ
- časopisecké články MeSH
- přehledy MeSH
Modelling the flow properties of rubber blends makes it possible to predict their rheological behaviour during the processing and production of rubber-based products. As the nonlinear nature of such complex processes complicates the creation of exact analytical models, it is appropriate to use artificial intelligence tools in this modelling. The present study was implemented to develop a highly efficient artificial neural network model, optimised using a novel training algorithm with fast parallel computing to predict the results of rheological tests of rubber blends performed under different conditions. A series of 120 real dynamic viscosity-time curves, acquired by a rubber process analyser for styrene-butadiene rubber blends with varying carbon black contents vulcanised at different temperatures, were analysed using a Generalised Regression Neural Network. The model was optimised by limiting the fitting error of the training dataset to a pre-specified value of less than 1%. All repeated calculations were made via parallel computing with multiple computer cores, which significantly reduces the total computation time. An excellent agreement between the predicted and measured generalisation data was found, with an error of less than 4.7%, confirming the high generalisation performance of the newly developed model.
- Klíčová slova
- curing process, generalised regression neural network, intelligent modelling, parallel computing, rubber blends,
- Publikační typ
- časopisecké články MeSH
This paper focuses on wrapper-based feature selection for a 1-nearest neighbor classifier. We consider in particular the case of a small sample size with a few hundred instances, which is common in biomedical applications. We propose a technique for calculating the complete bootstrap for a 1-nearest-neighbor classifier (i.e., averaging over all desired test/train partitions of the data). The complete bootstrap and the complete cross-validation error estimate with lower variance are applied as novel selection criteria and are compared with the standard bootstrap and cross-validation in combination with three optimization techniques - sequential forward selection (SFS), binary particle swarm optimization (BPSO) and simplified social impact theory based optimization (SSITO). The experimental comparison based on ten datasets draws the following conclusions: for all three search methods examined here, the complete criteria are a significantly better choice than standard 2-fold cross-validation, 10-fold cross-validation and bootstrap with 50 trials irrespective of the selected output number of iterations. All the complete criterion-based 1NN wrappers with SFS search performed better than the widely-used FILTER and SIMBA methods. We also demonstrate the benefits and properties of our approaches on an important and novel real-world application of automatic detection of the subthalamic nucleus.
- MeSH
- teoretické modely MeSH
- velikost vzorku * MeSH
- Publikační typ
- časopisecké články MeSH
- práce podpořená grantem MeSH
- validační studie MeSH
The separation process in capillary micellar electrochromatography (MEKC) can be modelled using artificial neural networks (ANNs) and optimisation of MEKC methods can be facilitated by combining ANNs with experimental design. ANNs have shown attractive possibilities for non-linear modelling of response surfaces in MEKC and it was demonstrated that by combining ANN modelling with experimental design, the number of experiments necessary to search and find optimal separation conditions can be reduced significantly. A new general approach for computer-aided optimisation in MEKC has been proposed which, because of its general validity, can also be applied in other separation techniques.
- MeSH
- chromatografie micelární elektrokinetická kapilární metody MeSH
- EDTA analogy a deriváty chemie MeSH
- kovy analýza MeSH
- neuronové sítě * MeSH
- počítačová simulace MeSH
- výzkumný projekt MeSH
- Publikační typ
- časopisecké články MeSH
- práce podpořená grantem MeSH
- Názvy látek
- CDTA MeSH Prohlížeč
- EDTA MeSH
- kovy MeSH
OBJECTIVES: This article summarized errors obtained by diverse techniques used for the derivation of cross-sectional contours in nonadult humeri and tibiae. MATERIALS AND METHODS: We analyzed cross-sectional contours in a total sample of 62 humeral and 75 tibial diaphyses in the age between birth and 12 years divided into three age groups. Long bone 35% (humeri) and midshaft (tibiae) cross-sections were taken on micro-CT images and analyzed by EPJMacro in FIJI. Properties were extracted from contours derived by manual, automatic, spline, and ellipse techniques. Agreement between techniques was assessed using manually extracted properties such as the true value using percent prediction error (%PE), reduced major axis regression, and ±95% limits of agreement. RESULTS: The lowest measurement errors were obtained for total areas, moderate errors for cortical areas and section moduli, and the highest errors for medullary areas for both bones. Derivation of humeral nonadult cross-sectional properties is less sensitive to the technique used for derivation of periosteal and endosteal contours, reaching mean %PEs below 5%. In contrast, tibial nonadult cross-sectional properties are more sensitive to the technique used and exceed 5% for some combinations. DISCUSSION: Automatic techniques provide reasonably high agreement with manually extracted contours for nonadult humeri but low agreement for tibiae. Semiautomatic approaches-spline and ellipse techniques-may reduce the error for all studied properties in tibiae, especially when combined with manually traced periosteal contours. The positive effect of the semiautomatic technique on measurement error is low for humeri.
- Klíčová slova
- EPJMacro, ImageJ, biomechanics, humerus, tibia,
- MeSH
- anatomie průřezová MeSH
- antropologie fyzická MeSH
- antropometrie MeSH
- dítě MeSH
- humerus * anatomie a histologie diagnostické zobrazování MeSH
- kojenec MeSH
- lidé MeSH
- novorozenec MeSH
- okostice * anatomie a histologie diagnostické zobrazování MeSH
- předškolní dítě MeSH
- rentgenová mikrotomografie MeSH
- reprodukovatelnost výsledků MeSH
- statistické modely MeSH
- tibie * anatomie a histologie diagnostické zobrazování MeSH
- Check Tag
- dítě MeSH
- kojenec MeSH
- lidé MeSH
- novorozenec MeSH
- předškolní dítě MeSH
- Publikační typ
- časopisecké články MeSH
- práce podpořená grantem MeSH
BACKGROUND: Infection with feline immunodeficiency virus (FIV) causes an immunosuppressive disease whose consequences are less severe if cats are co-infected with an attenuated FIV strain (PLV). We use virus diversity measurements, which reflect replication ability and the virus response to various conditions, to test whether diversity of virulent FIV in lymphoid tissues is altered in the presence of PLV. Our data consisted of the 3' half of the FIV genome from three tissues of animals infected with FIV alone, or with FIV and PLV, sequenced by 454 technology. RESULTS: Since rare variants dominate virus populations, we had to carefully distinguish sequence variation from errors due to experimental protocols and sequencing. We considered an exponential-normal convolution model used for background correction of microarray data, and modified it to formulate an error correction approach for minor allele frequencies derived from high-throughput sequencing. Similar to accounting for over-dispersion in counts, this accounts for error-inflated variability in frequencies - and quite effectively reproduces empirically observed distributions. After obtaining error-corrected minor allele frequencies, we applied ANalysis Of VAriance (ANOVA) based on a linear mixed model and found that conserved sites and transition frequencies in FIV genes differ among tissues of dual and single infected cats. Furthermore, analysis of minor allele frequencies at individual FIV genome sites revealed 242 sites significantly affected by infection status (dual vs. single) or infection status by tissue interaction. All together, our results demonstrated a decrease in FIV diversity in bone marrow in the presence of PLV. Importantly, these effects were weakened or undetectable when error correction was performed with other approaches (thresholding of minor allele frequencies; probabilistic clustering of reads). We also queried the data for cytidine deaminase activity on the viral genome, which causes an asymmetric increase in G to A substitutions, but found no evidence for this host defense strategy. CONCLUSIONS: Our error correction approach for minor allele frequencies (more sensitive and computationally efficient than other algorithms) and our statistical treatment of variation (ANOVA) were critical for effective use of high-throughput sequencing data in understanding viral diversity. We found that co-infection with PLV shifts FIV diversity from bone marrow to lymph node and spleen.
- MeSH
- algoritmy MeSH
- DNA virů genetika MeSH
- interpretace statistických dat * MeSH
- kočičí AIDS genetika imunologie virologie MeSH
- kočky MeSH
- nemoci koček genetika imunologie přenos virologie MeSH
- statistické modely * MeSH
- virus kočičí imunodeficience klasifikace genetika patogenita MeSH
- vysoce účinné nukleotidové sekvenování metody MeSH
- zvířata MeSH
- Check Tag
- kočky MeSH
- zvířata MeSH
- Publikační typ
- časopisecké články MeSH
- Research Support, N.I.H., Extramural MeSH
- Názvy látek
- DNA virů MeSH
