KU Leuven Research Data Repository
Not a member yet
662 research outputs found
Sort by
A Large and Diverse Ground Truth Dataset of Simulated Realistic Cardiac Ultrasound Recordings
No full textTo address the scarcity of cardiac ultrasound datasets crucial for developing and validating machine learning-based ultrasound data processing algorithms, we have created a ground truth dataset comprising 1296 synthetic cardiac ultrasound recordings. These recordings feature realistic speckle texture derived from clinical scans and left ventricular deformation (strain patterns) generated using the CircAdapt heart model. A detailed description of the dataset creation process is provided in the parent publication titled "Large-scale simulation of realistic cardiac ultrasound data with clinical appearance: methodology and open-access database".
In this dataset, each simulated recording includes ground truth left ventricular motion, beamformed radio frequency signals, scatter maps (comprising ultrasound scattering sites and the respective amplitudes), all for the entire cardiac cycle.
Specifically, in 204 out of the 1296 recordings, the left ventricular motion is imposed using regional longitudinal and radial strain patterns for two segments of the left ventricle: the left ventricular free wall (LVFW) and the septal wall (SPW). In the remaining 1092 recordings, left ventricular motion is imposed using both global longitudinal and radial strain patterns, as well as the average of the LVFW and SPW strain patterns.
This dataset serves as a valuable resource (e.g., a potential augmentation tool) for the development and validation of new machine and deep learning-based ultrasound data processing algorithms. This work is a part of MARie Curie Intelligent UltraSound (MARCIUS) project which is funded by the European Union’s Horizon 2020 research and innovation programme under the Marie Sklodowska Curie grant agreement number 860745
Mean Turbulence Statistics from Direct Numerical Simulations of Stably Stratified Channel Flows with One- and Two-Dimensional Surface Temperature Heterogeneity
No full textThe dataset contains a suite of Direct Numerical Simulation (DNS) results of pressure driven stably stratified channel flows, both with homogeneous and heterogeneous surface temperatures. The one-dimensional temperature heterogeneity is organized in streamwise infinite strips of width λy/2 with spanwise varying surface temperature with difference Δ θ, while the same temperature difference is maintained between the lower and upper wall to enforce overall stable stratification. In the cases with two-dimensional surface heterogeneity, the strips have a finite streamwise length λx/2.
Two different Reynolds numbers, i.e. Reτ=180 and Reτ=550, and two different Richardson numbers Riτ=120 and Riτ=960 are considered for the heterogeneous cases. For reference, a number of canonical stable channel flow cases with homogeneous surfaces at different Reynolds and Richardson numbers is also included. The simulations were run until a statistically steady state was reached, after which turbulence statistics were collected over a certain time window (30 h/uτ in most cases, see 'T_av' column in StableChannel_Overview.csv. ).
The dataset is generated with the SP-Wind code, an in-house DNS/LES code developed at KU Leuven. For details of the code structure and simulation set-up we refer to Bon and Meyers (2022) and Bon et al. (2023). We note that both publications also contain a schematic illustration of the surface heterogeneity configurations (Figure 1 there).
The dataset is organized as follows. A total of 46 simulations is included, and each has a separate folder. Each folder contains global simulation properties in flowproperties.csv, and horizontally-averaged profiles of relevant variables in profiles_cc.csv (variables on cell-centered grid) and profiles_st.csv (vertical derivatives on staggered grid). Besides, the folders of the streamwise homogeneous cases (kx = 0) contain streamwise-averaged yz-planes of velocity components, potential temperature, Reynolds stresses and temperature fluxes. For the cases with two-dimensional heterogeneity (kx > 0), phase-averaged 3D fields of these variables are included. Note that all variables are time-averaged. An overview of the mean flow properties of each simulation is provided in the file StableChannel_Overview.csv, which also includes relevant input parameters such as domain size and grid resolution.
Fore more information, we refer to the readme.txt file located in the dataset and to the two related publications.
Acknowledgements
The authors acknowledge support from the Research Foundation Flanders (FWO, Grant Number G098320N). The computational resources and services used in this work were provided by the VSC (Flemish Supercomputer Center), funded by the Research Foundation – Flanders (FWO) and the Flemish Government.
References
Bon, T. & Meyers, J. (2022), Stable channel flow with spanwise heterogeneous surface temperature. Journal of Fluid Mechanics (2022), vol. 933, A54, doi:10.1017/jfm.2021.1113
Bon, T., Broos, D., Cal, R. B., & Meyers, J. (2023), Secondary flows induced by two-dimensional surface temperature heterogeneity in stably stratified channel flow. Journal of Fluid Mechanics, vol. 970, A20, doi:10.1017/jfm.2023.619
</p
Replication Data for: High-resolution in-situ photo-irradiation MAS NMR: Application to the UV-polymerization of n butyl acrylate
No full textThis dataset contains NMR data files that were included in the peer-reviewed journal article: T.J.N.Hooper, R. de Oliveira-Silva, D. Sakellariou, High-resolution in-situ photo-irradiation MAS NMR: Application to the UV-polymerization of n butyl acrylate,
Journal of Materials Chemistry A, (under review).
A link to this dataset is provided in the Data Availability Statement of the article to adhere to the author guidelines of the Royal Society of Chemistry. The NMR data is used in the article to demonstrate a modified MAS NMR probe and windowed rotor that allows in-situ MAS NMR of samples under continuous irradiation. The 1H and 13C NMR data is used to follow the UV initiated polymerization reaction on n-butyl acrylate
Replication Data for: Utilizing gyroscopes for force identification in structural dynamics using Kalman filters
No full textRaw data for the paper of "Utilizing gyroscopes for force identification in structural dynamics using Kalman filters". The dataset contains the sensor readings for the experimental validation described in section 5 FE models used. The data can be directly accessed via MATLAB or GNU Octave (open access)
MixMatch: Flow Matching for Mixnet Traffic
No full textThis is the main repository for our PoPETs 2024.2 article "MixMatch: Flow Matching for Mixnet Traffic".
Mixnets provide communication anonymity against network adversaries by routing packets independently via multiple hops, delaying them artificially at each hop, and introducing cover traffic. We show that these features (particularly the use of cover traffic) significantly diminish the effectiveness of state-of-the-art flow correlation techniques developed to link the two ends of a Tor connection
Code of "On the Hardness of Probaiblistic Neurosymbolic Learning"
No full textCode to replicate all the experiments in the paper "On the Hardness of Probaiblistic Neurosymbolic Learning", published at ICML2024. The code can also be found at https://github.com/jjcmoon/hardness-nesy.
The limitations of purely neural learning have sparked an interest in probabilistic neurosymbolic models, which combine neural networks with probabilistic logical reasoning. As these neurosymbolic models are trained with gradient descent, we study the complexity of differentiating probabilistic reasoning. We prove that although approximating these gradients is intractable in general, it becomes tractable during training. Furthermore, we introduce WeightME, an unbiased gradient estimator based on model sampling. Under mild assumptions, WeightME approximates the gradient with probabilistic guarantees using a logarithmic number of calls to a SAT solver. Lastly, we evaluate the necessity of these guarantees on the gradient. Our experiments indicate that the existing biased approximations indeed struggle to optimize even when exact solving is still feasible
LMSD 2021 Dataset for Damage Identification in Plates
No full textLMSD 2021 Dataset for Damage Identification in Plates
These files contain data collected to experimentally validate damage identification methods for plate-like structures.
Experimental Setup
The inspected structure is a 600mm x 600mm x 4mm CFRP plate with crossply layup.
A 12 x 12 square grid of 144 nodes is defined on this plate, with 50mm spacing and 25mm offset from the edges of the plate.
The plate is hung with elastic bands to obtain free-free boundary conditions.
It is excited on its front-side with a PCB 086C03 impact hammer.
7 PCB 352a24 1D-accelerometers are installed on its back-side.
Additional masses are glued to the plate to reproduce the scattering effect of damage.
Two types of scenarios are considered: point-like masses of 55g at pre-defined nodes on the grid and an elongated mass of 315g covering several nodes.
A total of 6 damage scenarios are considered, 5 point-like scenarios with 1, 3, 4, 5 and 6 added masses, respectively, and 1 elongated-mass scenario with 1 added mass.
The included pictures summarize the experimental setup and show the 12 x 12 grid of nodes with the position of the 7 accelerometers/hammer impact locations in black and the positions of the added masses in red (the number labels at a given position indicate which scenarios have a mass added at that position).
Zoomed views are also provided of a 55g point-mass and a 315g elongated mass, glued to the plate.
Data Acquisition Procedure
To obtain the baseline FRFs, the responses are measured on the healthy plate from all 144 nodes to the 7 accelerometer nodes.
To obtain the damaged FRFs, the responses are measured on the damaged plate from the 7 accelerometer nodes to the 7 accelerometer nodes.
The resulting damaged responses form what is also referred to as the multistatic data matrix.
Each measurement is repeated 5 times and the responses are averaged to increase the SNR.
The measured FRFs consist of 8192 frequency bins between 0Hz and 1600Hz.
Data Structure and Organization
The baseline file contains:
frf 7x144x8192 double: baseline frf from all 144 nodes to the 7 accelerometer nodes.
frequency 1x8192 double [Hz]: the frequencies in Hertz.
node_position 144x2 double [mm]: position of the nodes in millimeters.
probing_nodes 1x7 uint16: indices of the accelerometer nodes, which are also the nodes that receive hammer excitation to collect the damaged responses.
Each damage scenario is associated to a file with a corresponding, descriptive filename. Each of the files contains:
frf 7x7x8192 double: damaged frf from the 7 accelerometer nodes to the 7 accelerometer nodes (also referred to as multistatic data matrix). By convention, frf(i, j, :) is the frf measured at accelerometer i when a hammer excitation is applied at the location of accelerometer j.
mass_nodes 1xn uint16: indices of the nodes at which a mass is added.
The datasets are provided both as mat files and npz files, alongside with helper functions to conveniently load the data and accelerate collaboration.
Note that the mat files use MATLAB-compatible indexing (i.e. node numbering starts at 1) while the npz files use conventional zero-based indexing (i.e. node numbering starts at 0).
For this reason all node indices in the probing_nodes and mass_nodes variables are offset by one in the mat files compared to the npz files.
Helper functions and examples demonstrating how to load and process the data are provided alongside this dataset under MIT Licensing.
Licensing
Data produced and made available by the LMSD group, KU Leuven, under CC BY 4.0 Licensing.
Code written by Nathan Dwek with the LMSD group, KU Leuven, made available under MIT Licensing.
Citing this Work
If you use the data itself directly, please cite this dataset appropriately. Proper citation text can be found on the dataset webpage in multiple formats.
Please cite the relevant publication(s) below if your work is based on or compares to the damage identification methods they introduce:
N. Dwek, V. Dimopoulos, D. Janssens, M. Kirchner, E. Deckers, and F. Naets, "Damage Identification in Plate-Like Structures Using Frequency-Coupled L1-Based Sparse Estimation," (Pre-print submitted to MSSP, October 31, 2023). Available at SSRN. doi: 10.2139/ssrn.4644311
N. Dwek, D. Janssens, M. Kirchner, and E. Deckers, "Sparse Damage Identification at Off-Grid Locations on Plate-like Structures using Frequency-Coupled Group Lasso," to be presented at the 2024 European Workshop on Structural Health Monitoring, June 2024.
Contact
Nathan Dwek - [email protected]
LMSD Group - [email protected]
Department of Mechanical Engineering
KU Leuven
Belgium
</p
Replication Data for: Investigation of the Temperature Effect on the Formation of a Two-Dimensional Self-Assembled Network at the Liquid/Solid Interface
No full textThe purpose of this data set is to give the reader access to the experimental data used in the publication "Investigation of the Temperature Effect on the Formation of a Two-Dimensional Self-Assembled Network at the Liquid/Solid Interface" ( DOI: ). This folder contains a dataset of scanning tunnelling microscopy (STM) images and quantitative analysis files used for a systematic study of the effect of concentration and temperature on the overall process of self-assembled molecular network formation at the liquid/solid interface for the case of (9H-fluoren-9-yl)methyl (2-(octadecylamino)-2-oxoethyl)carbamate at the 1-phenlyloctane/HOPG interface. Herein, the effect of the concentration and temperature on the adsorption behaviour, focusing on the surface coverage, has been evaluated at the nanoscale, and obtained results have been analysed using analytical thermodynamic models
Multi-Sensor Voice Command Dataset
No full textThe repository includes the audio files recorded from a wireless audio sensor network of 4 sensors. A total of 20 volunteers were instructed to repeat five English voice commands ("Lights On", "Lights Off", "Music On", "Music Stop", "Next Song") from three different positions, for a total of 15 recordings per keyword. Only in a few cases, the number of samples is reduced to 14 after manual data cleaning. Additionally, we registered 15 per-speaker generic spoken utterances, e.g., "Set an alarm to 7am", which are used as negative examples. Given a duration of 3 seconds per utterance, we collected a total of 1.5 hours of audio per sensor. Additionally, the negative data were augmented with 4.4 hours of recordings obtained by replaying the audio files from the test-clean and dev-clean sets of Librispeech using a set of speakers. The audio files are available from the weblink https://www.openslr.org/12 (dev-clean and test-clean repository), which are freely distributed under the CC-BY-4.0 license. Every recording was limited to 3 seconds, leading to a total amount of 5.9 hours of audio per sensor in our multi-sensor dataset.
This dataset favors the investigation of new speech recognition algorithms for audio data recorded with a network of ultra-low-power smart audio sensors. The application is voice command recognition, also known as keyword spotting, where the algorithm must recognize a voice command (e.g. "Lights On"), distinguishing it from other voice commands or other audio tracks, i.e. the negative data. Within a wireless audio sensor network scenario, the speech recognition algorithms are fed with the audio data recorded from multiple sensors, which are located in the environment
TRNG Entropy Model in the Presence of Flicker FM Noise
No full textThis repository provides a reference implementation for assessing the entropy contribution of flicker FM noise in an Elementary Ring Oscillator (ERO) True Random Number Generator (TRNG). A Monte Carlo simulation script is available to estimate the conditional entropy, given knowledge of the TRNG state. Additionally, it also includes figure generation scripts to visualize the simulated model data