256869 research outputs found
Sort by
Score-based diffusion models for diffuse optical tomography with uncertainty quantification
No full textScore-based diffusion models are a recently developed framework for posterior sampling in Bayesian inverse problems with a state-of-the-art performance for severely ill-posed problems by leveraging a powerful prior distribution learned from empirical data. Despite generating significant interest especially in the machine-learning community, a thorough study of realistic inverse problems in the presence of modelling error and utilization of physical measurement data is still outstanding. In this work, the framework of unconditional representation for the conditional score function (UCoS) is evaluated for linearized difference imaging in diffuse optical tomography (DOT). DOT uses boundary measurements of near-infrared light to estimate the spatial distribution of absorption and scattering parameters in biological tissues. The problem is highly ill-posed and thus sensitive to noise and modelling errors. We introduce a novel regularization approach that prevents overfitting of the score function by constructing a mixed score composed of a learned and a model-based component. Validation of this approach is done using both simulated and experimental measurement data. The experiments demonstrate that a data-driven prior distribution results in posterior samples with low variance, compared to classical model-based estimation, and centred around the ground truth, even in the context of a highly ill-posed problem and in the presence of modelling errors
Predicting the Unseen: Transductive Transfer Learning in Real Estate Price Prediction
No full textThis study examines transfer learning for time-dependent newly built apartment price prediction using spatial features only, motivated by their higher temporal stability compared to sociodemographic or economic variables. We evaluate whether a model trained in data-rich settings can generalize to unseen areas and to a different city. In Vienna, the transfer setup achieved a mean absolute percentage error (MAPE) of 20%–30% for 1-year predictions in unseen areas, with a maximum negative error (transfer-induced performance loss relative to the non-transfer baseline) of 11%, and performed particularly well in developing areas (MAPE ≈10%). When transferred to another city, the maximum negative error remained below 8% under complete spatial feature coverage, with a minimum MAPE of 2.08% for 1-year predictions. Finally, reducing the spatial feature set from 118 to approximately half preserved predictive performance while lowering computational complexity and improving practical applicability. Overall, the results suggest that spatial feature-based transfer learning can provide competitive short-term accuracy for apartment price prediction while supporting principled feature reduction
Competing instabilities for models with non-local and retarded interactions: a functional renormalization group perspective
No full text
Quality-controlled deformation analysis of the 26-m HartRAO radio telescope's main reflector: First results
No full textRadio telescopes are pivotal in receiving radio frequencies from space. These telescopes, typically featuring parabolic dishes, focus radio waves onto a central receiving point to amplify the incoming signal. The stability of the telescope’s main reflector’s shape across various orientations is crucial, as deformations can distort the received signal. This study focuses on the 26-meter radio telescope at the Hartebeesthoek Radio Astronomy Observatory (HartRAO) in South Africa. A high-end laser scanner is employed to record the surface of the rotating paraboloid reflector in multiple orientations. The telescope is capable of moving through different declinations and hour angles requiring to measuring 88 different positions of the telescope, to provide a complete
picture of the deformations. Fitting models are applied to estimate the shape of the rotating paraboloid from the raw data also considering calibration errors of the laser scanner used. First results for deformation patterns and, therefore, the local deformations of the main reflector are shown
Optimal Current Trajectory Evaluation for Sensorless Controlled Synchronous Machines based on Finite Element Analysis
No full textThe performance of sensorless controlled synchronous drives is strongly depending on the electromagnetic and mechanical design of the synchronous machine itself. This paper discusses a methodology to analyze the self-sensing capability of different synchronous machine types to evaluate an optimized operational strategy by a special current trajectory for low-speed and standstill especially with respect to sensorless operation. A comprehensive analysis of the machine's differential inductances by using finite element simulation enables the characterization of achievable sensorless rotor position accuracy by the help of statistical properties. A sensorless error angle map is generated over a certain area around the intended operation range for each individual machine respectively. Depending on specific operational criteria a new optimal current trajectory for sensorless control is developed replacing commonly used maximum torque per ampere (MTPA) strategy, which usually does not take any specific self-sensing capability into account. The paper shows the proposed evaluation process applied to three synchronous machines exemplarily
Adaptive Power Control for Laser Triangulation Measurements on Moving Samples
No full textThis article proposes an adaptive laser power control approach for laser triangulation measurements on moving samples. Utilizing the obtained image intensities as a feedback signal, the effective sample reflectivity is estimated. Based on this estimation, a digital integral controller is adaptively tuned. The thereby determined laser power setpoint is stabilized by a custom laser driver with a bandwidth of 50 kHz. Implementing the digital controller on an FPGA enables power adjustments between two consecutive exposure times, ensuring stable laser power conditions during the exposure. Experimental evaluation on a sloped sample successfully demonstrates that the proposed method reduces the spot size variation on the detector by 81%, ensuring a constant spot size for arbitrary sample orientations and exposure times