Compression Scenario Research Articles

Using Highly Compressed Gradients in Federated Learning for Data Reconstruction Attacks

Federated learning (FL) preserves data privacy by exchanging gradients instead of local training data. However, these private data can still be reconstructed from the exchanged gradients. Deep leakage from gradients (DLG) is a classical reconstruction attack that optimizes dummy data to real data by making the corresponding dummy and real gradients as similar as possible. Nevertheless, DLG fails with highly compressed gradients, which are crucial for communication-efficient FL. In this study, we propose an effective data reconstruction attack against highly compressed gradients, called highly compressed gradient leakage attack (HCGLA). In particular, HCGLA is characterized by the following three key techniques: 1) Owing to the unreasonable optimization objective of DLG in compression scenarios, we redesign a plausible objective function, ensuring that compressed dummy gradients are similar to the compressed real gradients. 2) Instead of simply initializing dummy data through random noise, as in DLG, we design a novel dummy data initialization method, Init-Generation, to compensate for information loss caused by gradient compression. 3) To further enhance reconstruction quality, we train an ad hoc denoising model using the methods of “first optimizing, next filtering, and then reoptimizing”. Extensive experiments on various benchmark data sets and mainstream models show that HCGLA is an effective reconstruction attack even against highly compressed gradients of 0.1%, whereas state-of-the-art attacks can only support 70% compression, thereby achieving a 700-fold improvement.

Application of Vertical Electrical Sounding in Mapping Lateral and Vertical Changes in the Subsurface Lithologies: A Case Study of Olbanita, Menengai Area, Nakuru, Kenya

Much study has been done in the study area linking Vertical Electrical Sounding (VES) interpreted results to lithologies in the subsurface though only tend to indicate the vertical changes with the aim of mapping the occurrence of groundwater aquifers. Several boreholes have been drilled in the study area, though not much has been done to compare the vertical and lateral lithologic changes in the study area. This research is based on VES modelled geoelectric layers compared from point to point and using borehole logs as control data to establish inferences of certain lithology in the subsurface. The inversion of each VES curve was obtained using an AGI Earth Imager ID inversion automated computer program and resistivities and thicknesses of a geoelectric model were estimated. The analyzed VES data interpretation achieved using the curve matching technique resulted in mapping the subsurface of the area as portraying H-type; ρ1 > ρ2 ρ3, K-type; ρ1 ρ2 > ρ3, A-type; ρ1 ρ2 ρ3, Q-type; ρ1 > ρ2 > ρ3, representing 3-Layer subsurface and subsequently a combination of HK, HA and KHK types of curves representing 4-Layer and 5-Layer in the subsurface. The analysis further deployed the use of the surfer software capabilities which combined the VES data to generate profiles running in the west-east and the north-south direction. A closer analysis of the curve types indicates that there exists a sequence showing a shifting of the order of arrangement between the west and the east fragments which incidentally coincides with VES points 8, 9 and 10 in the West-East profiles. The lateral change is noted from the types of curves established and each curve indicates a vertical change in the subsurface. Control log data of lithologies from four boreholes BH1, BH2, BH3 and BH5 to show a qualification that different resistivity values portent different lithologies. Indeed, an analysis at borehole BH3 lithologies is dominated by either compacted rocks or soils, insinuating a scenario of compression experienced in this part of the subsurface which confirmed compression of subsurface formations. A correlation of the VES curve types and their change from one point to another in the study area are evident. This change supported by the surfer generated profiles from the modeled VES data show that there exists and inferred fault line running in the north-south in the area. The inferred fault line by VES mapping, is magnificently outlined by the geological map. There is exuded evidence from this study that the application of VES is able to help map the lateral and the vertical changes in the subsurface of any area but the evidence of the specific lithologies has to be supported by availability of borehole log control data. The VES data was able to enumerate vertical layering of lithologies, lateral changes and even mapping vertical fault line in the study area.

Compression scenarios for Federated Learning in Smart Manufacturing

Recent advances in Industrial Internet of Things (IIoT) and communication technologies have provided new concepts of smart manufacturing and paved the way for the new era of Industry 4.0. The huge amount of generated data from machinery, connected systems, and edge devices can be exploited to extract actionable insights that can enhance the decision-making process in smart manufacturing, targeting to improve the return of investment. Conventionally, smart decisions are obtained from centralised data using predefined machine/deep learning models. To achieve better accuracy, these models require training based on diverse, unbiased, and comprehensive datasets, which may not be feasible in distributed and remote facilities due to the high scalability of smart manufacturing and data privacy concerns. Federated learning stands as a primary solution to these issues by bringing collaborative smart manufacturing without requiring centralised data training rather than model sharing. However, these models with many parameters result in big size, which put pressure on the communication channels due to frequent transfers from server to clients and vice versa and may face communication delays and model corruption. In addition, bigger models create issues for edge devices with limited memory and processing capacity which needs consideration. Therefore, effective compression methods are mandatory to ensure Communication-Efficient Federated Learning. In this paper, first, we discuss some possible application opportunities of FL in smart manufacturing, especially for the automotive industry but that can be extended for others. Secondly, we highlight FL communication challenge in smart manufacturing for which we present the compression concept and objectives in FL. It is followed by discussing the state-of-the-art compression techniques in FL. Thirdly, a special focus is given to NNR and DeepCABAC for their demonstrated high compression gains with a wide variety of deep models. Finally, we gave some directions for compressing the FL models in the future.

Underwater localization and node mobility estimation

<p><span lang="EN-US">In this paper, localizing a moving node in the context of underwater wireless sensor networks (UWSNs) is considered. Most existing algorithms have had designed to work with a static node in the networks. However, in practical case, the node is dynamic due to relative motion between the transmitter and receiver. The main idea is to record the time of arrival message (ToA) stamp and estimating the drift in the sampling frequency accordingly. It should be emphasized that, the channel conditions such as multipath and delay spread, and ambient noise is considered to make the system pragmatic. A joint prediction of the node mobility and speed are estimated based on the sampling frequency offset estimation. This sampling frequency offset drift is detected based on correlating an anticipated window in the orthogonal frequency division multiplexing (OFDM) of the received packet. The range and the distance of the mobile node is predicted from estimating the speed at the received packet and reused in the position estimation algorithm. The underwater acoustic channel is considered in this paper with 8 paths and maximum delay spread of 48 ms to simulate a pragmatic case. The performance is evaluated by adopting different nodes speeds in the simulation in two scenarios of expansion and compression. The results show that the proposed algorithm has a stable profile in the presence of severe channel conditions. Also, the result shows that the maximum speed that can be adopted in this algorithm is 9 km/h and the expansion case profile is more stable than the compression scenario. In addition, a comparison with a dynamic triangular algorithm (DTN) is presented in order to evaluate the proposed system.</span></p>

Priority‐based encoding of triangle mesh connectivity for a known geometry

AbstractIn certain practical situations, the connectivity of a triangle mesh needs to be transmitted or stored given a fixed set of 3D vertices that is known at both ends of the transaction (encoder/decoder). This task is different from a typical mesh compression scenario, in which the connectivity and geometry (vertex positions) are encoded either simultaneously or in reversed order (connectivity first), usually exploiting the freedom in vertex/triangle re‐indexation. Previously proposed algorithms for encoding the connectivity for a known geometry were based on a canonical mesh traversal and predicting which vertex is to be connected to the part of the mesh that is already processed. In this paper, we take this scheme a step further by replacing the fixed traversal with a priority queue of open expansion gates, out of which in each step a gate is selected that has the most certain prediction, that is one in which there is a candidate vertex that exhibits the largest advantage in comparison with other possible candidates, according to a carefully designed quality metric. Numerical experiments demonstrate that this improvement leads to a substantial reduction in the required data rate in comparison with the state of the art.

Characterization of a Non-Darcy Flow and Development of New Correlation of NON-Darcy Coefficient

Darcy’s law has long been used to describe the flow in porous media. Despite the progress that took place in oil production industry research, it became clear that there is a loss of pressure, especially in the area near the wellbore region, where Darcy’s law is not applicable. For this reason, Forchheimer presented his equation in 1910, where he added a new term to Darcy’s law dealing with pressure loss due to inertial forces by introducing a new term, the β coefficient, into the equation. This paper presents a study of fluid flow through porous media, where water was used as a working fluid. Furthermore, the characteristics of the non-Darcy flow were analyzed by presenting the corresponding pressure and velocity gradient curves for each pressure. Extensive analysis indicates that many of the correlations available in the literature either have defective units or are the product of a small number of experiments. In this study, we benefit from relatively large samples, the radial flow, and the perforation in the middle of the samples. The properties of the samples were measured using mercury intrusion porosimetry. It was found that there is a direct relationship between the porosity and the grain’s size; the greater the size of the grains, the greater the porosity, and vice versa. The non-Darcy coefficient term, β, is found to be inversely proportional to the porosity and permeability. In a previous study, the β was investigated for compressible flow scenarios; however, this study calculated it for an incompressible flow. Finally, by analyzing the β values of both studies, we could deduce new novelty correlations for the β coefficient term, where the permeability, porosity, and tortuosity are included.

Probing compressed higgsinos with forward protons at the LHC

Supersymmetric models with nearly-degenerate light higgsinos provide a consistent solution to the naturalness problem under rigorous constraints from current experimental searches for sparticles. However, it is challenging to probe the compressed scenarios at collider experiments due to the hard-to-detect soft final states. To overcome this issue, strategies have been proposed to take advantage of the photon fusion along the ultraperipheral collision at the Large Hadron Collider, which are feasible with the forward detectors installed at the ATLAS and CMS detectors. In this paper, we demonstrate a search strategy for the chargino pair production via photon fusion at the 13 TeV LHC, realizing a good sensitivity that can exclude {m}_{{overset{sim }{chi}}_1^{pm }} up to 270 GeV (308 GeV) for a mass splitting ∆m( {overset{sim }{chi}}_1^{pm } , {overset{sim }{chi}}_1^0 ) ∈ [1, 15] GeV with an integrated luminosity of 100 fb−1 (3 ab−1) at 95% C.L.

Using the polarization properties of double radio relics to probe the turbulent compression scenario

Context. Radio relics are megaparsec-sized synchrotron sources located in the outskirts of some merging galaxy clusters. Binary-merging systems with a favorable orientation may host two almost symmetric relics, named double radio relics. Aims. Double radio relics are seen preferentially edge-on and, thus, constitute a privileged sample for statistical studies. Their polarization and Faraday rotation properties give direct access to the relics’ origin and magnetic fields. Methods. In this paper, we present a polarization and rotation measure (RM) synthesis study of four clusters hosting double radio relics, namely 8C 0212+703, Abell 3365, and PLCK G287.0+32.9; previously missing polarization studies; and ZwCl 2341+0000, for which conflicting results have been reported. We used 1–2 GHz Karl G. Jansky Very Large Array observations. We also provide an updated compilation of known double radio relics with important observed quantities. We studied their polarization and Faraday rotation properties at 1.4 GHz and we searched for correlations between fractional polarization and physical resolution, the distance from the cluster center, and the shock Mach number. Results. The weak correlations found between these quantities are well reproduced by state-of-the-art magneto-hydrodynamical simulations of radio relics, confirming that merger shock waves propagate in a turbulent medium with tangled magnetic fields. Both external and internal Faraday depolarization should play a fundamental role in determining the polarization properties of radio relics at 1.4 GHz. Although the number of double radio relics with RM information is still low, their Faraday rotation properties (i.e., rest-frame RM and RM dispersion below 40 rad m−2 and non-Gaussian RM distribution) can be explained in the scenario in which shock waves with Mach numbers larger than 2.5 propagate along the plane of the sky and compress the turbulent intra-cluster medium.

Molecular Gas Structures Traced by 13CO Emission in the 18,190 12CO Molecular Clouds from the MWISP Survey

After morphological classification of 18,190 12CO molecular clouds, we further investigate the properties of their internal molecular gas structures traced by the 13CO (J = 1−0) line emissions. Using three different methods to extract the 13CO gas structures within each 12CO cloud, we find that ∼15% of the 12CO clouds (2851) have 13CO gas structures and these 12CO clouds contribute about 93% of the total integrated flux of 12CO emission. In each of the 2851 12CO clouds with 13CO gas structures, the 13CO emission area generally does not exceed 70% of the 12CO emission area, and the 13CO integrated flux does not exceed 20% of the 12CO integrated flux. We reveal a strong correlation between the velocity-integrated intensities of 12CO lines and those of 13CO lines in both 12CO and 13CO emission regions. This indicates the H2 column densities of molecular clouds are crucial for the 13CO line emission. After linking the 13CO structure detection rates of the 18,190 12CO molecular clouds to their morphologies, i.e., nonfilaments and filaments, we find that the 13CO gas structures are primarily detected in 12CO clouds with filamentary morphologies. Moreover, these filaments tend to harbor more than one 13CO structure. That demonstrates filaments not only have larger spatial scales, but also have more molecular gas structures traced by 13CO lines, i.e., local gas density enhancements. Our results favor the turbulent compression scenario for filament formation, in which dynamical compression of turbulent flows induces local density enhancements. The nonfilaments tend to be in the low-pressure and quiescent turbulent environments of the diffuse interstellar medium.

Numerical study of a liquid-piston compressor system for hydrogen applications

The use of a liquid-piston system for hydrogen compression is investigated in this paper by means of a computational fluid dynamics (CFD) analysis. In the specific context of hydrogen-driven vehicles, high-pressure storage tanks are key to provide substantial range. The present study focuses on the intermediary compression stage of a compression-storage-dispensing (CSD) station, bringing hydrogen gas from 15 bar to 450 bar, i.e., for a pressure ratio of 30. Until now the liquid-piston technology has not been investigated for hydrogen gas compression at very high pressure, which is the purpose of this study. Simulations of the compressible two-phase flow problem are performed with a volume-of-fluid (VOF) framework using a real gas model for the gaseous phase to account for compressibility effects at large pressure ratios. A particular attention is paid to the numerical model formulation and to the treatment of the thermal boundary conditions. Results are reported using both time-resolved instantaneous bulk thermodynamic variables and global integrated quantities. Different compression scenarios are investigated, which highlights the compromise between compression efficiency and power density. To achieve the targeted pressure ratio at a power density of approximately 540 kW/m3, the compression energy cost reaches 1.67 kWh/kg. Finally the paper proposes an innovative solution to minimise cost and achieve quasi-isothermal compression, based on internal forced convection. For a similar power density, a high-speed fan in the top part of the compression chamber (modelled as a volumetric momentum source of 2500 N/m3) increases heat transfer and leads to a 25-% reduction in compression consumption.

Locality-based transfer learning on compression autoencoder for efficient scientific data lossy compression

Scientific simulation can generate petabyte-level data per run nowadays. To significantly reduce the data size while simultaneously maintaining the compression quality based on certain user requirements, error-bounded lossy compression techniques such as SZ and ZFP are now becoming popular. However, these techniques still cannot achieve a reduction ratio of more than two orders of magnitude with a low compression error. On the other hand, in deep learning, the autoencoder techniques have been widely used in data compression, especially images. As an alternative, the compression autoencoder (CAE) has recently been investigated to compress the scientific data. Although CAE provides a higher compression ratio than SZ and ZFP, it suffers from a high training overhead, which makes it almost impractical in real compression scenarios. In this paper, we propose a new locality-based transfer learning method in order to significantly increase the training speed of CAE while achieving a high compression ratio. We also adopt incremental learning to maintain a high prediction accuracy and use KL-divergence as an indicator to quickly identify whether a target domain has a low testing error. Our evaluation results show that, after using the locality-based transfer learning, the training time can be reduced by up to 1200 times, and still has a 2 to 4X compression ratio gain over the state-of-the-art scientific data lossy compressor SZ.

Comparison of Jaw-Thrust Maneuver and Standard Method for Airway Management with Laryngeal Mask Airway by Paramedics during Chest Compression: A Randomized, Crossover, Manikin Study.

The first priority of the primary survey of trauma care is airway management. For patients who have a known or suspected cervical spine injury, using the jaw-thrust maneuver is critical. It was hypothesized that the jaw-thrust maneuver would ease the insertion of the laryngeal mask airway (LMA) by moving the tongue forward from the palate and posterior pharyngeal wall. The aim of the study was to evaluate the effect of jaw-thrust maneuver on LMA insertion times of the paramedics with or without chest compression and with or without cervical stabilization in a manikin. Eleven experienced paramedics inserted LMA in jaw-thrust position and standard position in chest compression without cervical stabilization scenario, chest compression with cervical stabilization scenario, cervical stabilization without chest compression scenario, and the scenario where neither cervical stabilization nor chest compression were performed. The primary outcome of the study was the comparison of LMA insertion times for each method. The secondary outcome measures were first-pass success rates and the comparison of the difficulty level of each method. During the LMA placement, performing the jaw-thrust maneuver instead of the standard method did not shorten the LMA insertion times. Adding chest compression and/or cervical stabilization did not complicate the LMA insertion. All of the LMA insertion attempts during the jaw-thrust maneuver and standard method were successful. The findings of this study suggest that LMA insertion might be attempted both during the jaw-thrust maneuver and standard position in patients with or without chest compression and with or without cervical stabilization.

Optimized Load-Independent Hyperelastic Microcharacterization of Human Brain White Matter

Abstract A micromechanical methodology combined with genetic algorithm (GA) as a global optimization method is used to find the material properties of axons and extracellular matrix (ECM) in corpus callosum which is a part of human brain white matter. Studies have shown that axons are highly oriented in the ECM which enables us to approximate brain white matter as a unidirectional fibrous composite model. Using the one-term Ogden hyperelastic constitutive equations for the constituents and knowing the mechanical response of corpus callosum, GA optimization procedure is used in conjunction with finite element (FE) micromechanical analysis to find optimal material parameters for axon and ECM in three uniaxial loading scenarios of tension, compression, and simple shear. Moreover, by simultaneous fitting to the three loading modes' responses and applying Nelder–Mead simplex optimization method, best-fit parameters are found. The best-fit parameters can be used to approximate the behavior of axons and ECM in different uniaxial loading conditions with the minimum error and hence, can be interpreted as load-independent parameters. Micromechanical simulations by best-fit parameters show maximum stress increase of 2% and 29% for tension and shear and less than 1% reduction for compression mode compared to the case where optimal parameters are used. The findings and the methodology of this study can be employed for constitutive modeling of axonal fibers and its implementation in human head FE model where load-independent parameters are needed for simulating different loading scenarios.

On the Polarisation of Radio Relics

Radio relics are extended radio emission features which trace shock waves in the periphery of galaxy clusters originating from cluster mergers. Some radio relics show a highly polarised emission, which make relics an excellent probe for the magnetisation of the intra-cluster medium. The origin of the relic polarisation is still debated. It could be a result of tangentially stretching the magnetic field at the shock surface. This scenario would naturally explain the alignment of the polarisation (E-vectors) with the shock normal. We have implemented a toy model for the relic polarisation according to this scenario. We find that the magnetic field strength itself crucially affects the fractional polarisation. Moreover, we find that the shock strength has surprisingly little effect on the overall polarisation fraction. Finally, we find that the fractional polarisation may decrease downstream depending on the magnetic field strength. Our results demonstrates that the shock compression scenario provides a very plausible explanation for the radio relic polarisation which specific features permitting to test the origin of radio relic polarisation.

An Experimental Investigation of the Gas Permeability of Tectonic Coal Mineral under Triaxial Loading Conditions

Underground coal mining of CH4 gas-rich tectonic coal seams often induces methane outburst disasters. Investigating gas permeability evolution in pores of the tectonic coal is vital to understanding the mechanism of gas outburst disasters. In this study, the triaxial loading–unloading stresses induced gas permeability evolutions in the briquette tectonic coal samples, which were studied by employing the triaxial-loading–gas-seepage test system. Specifically, effects of loading paths and initial gas pressures on the gas permeability of coal samples were analyzed. The results showed the following: (1) The gas permeability evolution of coal samples was correlated with the volumetric strain change during triaxial compression scenarios. In the initial compaction and elastic deformation stages, pores and cracks in the coal were compacted, resulting in a reduction in gas permeability in the coal body. However, after the yield stage, the gas permeability could be enhanced due to sample failure. (2) The gas permeability of the tectonic coal decreased as a negative exponential function with the increase in initial gas pressure, in which the permeability was decreased by 67.32% as the initial gas pressure increased from 0.3 MPa to 1.5 MPa. (3) Coal samples underwent a period of strain development before they began to fail during confining pressure releasing. After the stress releasing-induced yield stage, the coal sample was deformed and cracked, resulting in a quickly increase in gas permeability. With a further releasing process, failure of the sample occurred, and thus induced rapidly increasing gas permeability. These obtained results could provide foundations for gas outburst prevention in mining gas-rich tectonic coal seams.

Probing Compressed Slepton Production at Future Large Hadron Collider

Various supersymmetry searches are carried out at the Large Hadron Collider (LHC) at CERN. While some searches have focused on the color sector, others focused on electroweak production of charginos-neutralinos and sleptons. Since no sign of supersymmetry has yet to be found, limits are set on the masses of the supersymmetric particles. In this phenomenological study, we probe direct slepton signal production in compressed mass spectra scenario at the potential future proton-proton collider operating at a center of mass energy of 33 TeV. We show that slepton signal with masses up to 270 GeV can be discovered at the future hadron collider with the data corresponding to 100 fb-1 integrated luminosity when ΔM =5 GeV.

An end-to-end deep learning model for robust smooth filtering identification

Smooth filtering, a common blurring and denoising operator, has often been utilized postoperatively to diminish the traces left by malicious manipulations. Most of the existing forensic methods only focus on one specific filtering artifact such as median filtering, which is insufficient to reveal the manipulation history of digital images. Unlike traditional convolutional neural network (CNN)-based networks, which normally introduce handcrafted features, including frequency domain features and median filtering residuals, into the preprocessing layer, this paper proposes an end-to-end deep learning model for robust smooth filtering identification. First, a distinctive network structure named the Squeeze-and-Excitation (SE) block is introduced to select discriminative features adaptively and suppress the irrelevant features to the smooth filtering effect. Then, as the network depth increases, multiple inception-residual blocks are stacked to extract discriminative features and reduce the information loss. Finally, different smooth filtering operations can be classified through learning hierarchical features. The experimental results on a composite database show that the proposed model outperforms the state-of-the-art methods, especially in small size and JPEG compression scenarios.

Comprehensive Operation Risk Assessment of a Highway Maintenance Area Based on Reliability

To study the influence of various working conditions on traffic safety in the maintenance area of a mountain expressway, 675 groups of PTV VISSIM traffic simulation experiments were designed with various traffic organization modes, traffic volumes, traffic compositions, activity area lengths, and speed limit values. The results show that the activity area length of a closed lane, a compressed lane, and a borrowed opposite lane had no marked influence on the traffic conflicts. There was a significant positive correlation between the proportion of trucks and the number of traffic conflicts, and the number of traffic conflicts increased significantly with an increase in traffic volume. In the closed lane and borrowed opposite lane scenarios, the increase in traffic conflicts was more obvious with the increase in traffic volume. There were obvious differences in the number of traffic conflicts under different forms of traffic organization. The number of conflicts in the compressed lane scenario was the lowest, and in the borrowed opposite lane scenario the number of conflicts was the highest. There was a significant correlation between a decrease in the speed limit and an increase in the number of traffic conflicts. Finally, with traffic volume, truck proportion, and speed limit values as independent variables and reliability as a dependent variable, linear regression equations of reliability were established for three traffic organization scenarios: closed lane, borrowed opposite lane, and compressed lane.

Nonlinear multiscale simulation of elastic beam lattices with anisotropic homogenized constitutive models based on artificial neural networks

A sequential nonlinear multiscale method for the simulation of elastic metamaterials subject to large deformations and instabilities is proposed. For the finite strain homogenization of cubic beam lattice unit cells, a stochastic perturbation approach is applied to induce buckling. Then, three variants of anisotropic effective constitutive models built upon artificial neural networks are trained on the homogenization data and investigated: one is hyperelastic and fulfills the material symmetry conditions by construction, while the other two are hyperelastic and elastic, respectively, and approximate the material symmetry through data augmentation based on strain energy densities and stresses. Finally, macroscopic nonlinear finite element simulations are conducted and compared to fully resolved simulations of a lattice structure. The good agreement between both approaches in tension and compression scenarios shows that the sequential multiscale approach based on anisotropic constitutive models can accurately reproduce the highly nonlinear behavior of buckling-driven 3D metamaterials at lesser computational effort.

Visual attention during pediatric resuscitation with feedback devices: a randomized simulation study

BackgroundThe aim of this study was to investigate the effect of feedback devices on visual attention and the quality of pediatric resuscitation.MethodsThis was a randomized cross-over simulation study at the Medical University of Vienna. Participants were students and neonatal providers performing four resuscitation scenarios with the support of feedback devices randomized. The primary outcome was the quality of resuscitation. Secondary outcomes were total dwell time (=total duration of visit time) on areas of interest and the workload of participants.ResultsForty participants were analyzed. Overall, chest compression (P < 0.001) and ventilation quality were significantly better (P = 0.002) when using a feedback device. Dwell time on the feedback device was 40.1% in the ventilation feedback condition and 48.7% in the chest compression feedback condition. In both conditions, participants significantly reduced attention from the infant’s chest and mask (72.9 vs. 32.6% and 21.9 vs. 12.7%). Participants’ subjective workload increased by 3.5% (P = 0.018) and 8% (P < 0.001) when provided with feedback during a 3-min chest compression and ventilation scenario, respectively.ConclusionsThe quality of pediatric resuscitation significantly improved when using real-time feedback. However, attention shifted from the manikin and other equipment to the feedback device and subjective workload increased, respectively.ImpactCardiopulmonary resuscitation with feedback devices results in a higher quality of resuscitation and has the potential to lead to a better outcome for patients.Feedback devices consume attention from resuscitation providers.Feedback devices were associated with a shift of visual attention to the feedback devices and an increased workload of participants.Increased workload for providers and benefits for resuscitation quality need to be balanced for the best effect.