multiphysics finite element simulation results of the capacitor model  (COMSOL Inc)

Journal: Frontiers in Artificial Intelligence

Article Title: Impact on bias mitigation algorithms to variations in inferred sensitive attribute uncertainty

doi: 10.3389/frai.2025.1520330

Figure Lengend Snippet: Overall design measuring the impact of uncertain sensitive attribute inference on bias mitigation algorithms.

Article Snippet: For the COMPAS and credit card client data sets, we use the simulated sensitive attribute inference model results with random misclassification at 0.75 balanced accuracy.

Techniques:

Journal: Frontiers in Artificial Intelligence

Article Title: Impact on bias mitigation algorithms to variations in inferred sensitive attribute uncertainty

doi: 10.3389/frai.2025.1520330

Figure Lengend Snippet: Gender inference accuracy using existing demographic inference models.

Article Snippet: For the COMPAS and credit card client data sets, we use the simulated sensitive attribute inference model results with random misclassification at 0.75 balanced accuracy.

Techniques:

Journal: Frontiers in Artificial Intelligence

Article Title: Impact on bias mitigation algorithms to variations in inferred sensitive attribute uncertainty

doi: 10.3389/frai.2025.1520330

Figure Lengend Snippet: Balanced accuracy and fairness score of the outcome prediction model (using inferred sensitive attribute) on Wikidata.

Article Snippet: For the COMPAS and credit card client data sets, we use the simulated sensitive attribute inference model results with random misclassification at 0.75 balanced accuracy.

Techniques:

Journal: Frontiers in Artificial Intelligence

Article Title: Impact on bias mitigation algorithms to variations in inferred sensitive attribute uncertainty

doi: 10.3389/frai.2025.1520330

Figure Lengend Snippet: Prediction model fairness difference using the ground truth sensitive attribute S and the inferred sensitive attribute S ′ with 0.75 balanced accuracy.

Article Snippet: For the COMPAS and credit card client data sets, we use the simulated sensitive attribute inference model results with random misclassification at 0.75 balanced accuracy.

Techniques:

Journal: Frontiers in Artificial Intelligence

Article Title: Impact on bias mitigation algorithms to variations in inferred sensitive attribute uncertainty

doi: 10.3389/frai.2025.1520330

Figure Lengend Snippet: Prediction model fairness difference between baseline model and bias mitigation methods using inferred sensitive attribute with 0.75 balanced accuracy.

Article Snippet: For the COMPAS and credit card client data sets, we use the simulated sensitive attribute inference model results with random misclassification at 0.75 balanced accuracy.

Techniques:

Journal: Microsystems & Nanoengineering

Article Title: A low-voltage-driven MEMS ultrasonic phased-array transducer for fast 3D volumetric imaging

doi: 10.1038/s41378-024-00755-9

Figure Lengend Snippet: a – c Normalized vibration velocities of all cells within one single element characterized by equivalent circuit model (EQC) simulations, COMSOL simulations and laser Doppler velocimetry (LDV) measurements. a Equivalent circuit model (EQC) simulation results. b Finite element method (FEM) simulation results. c Laser Doppler velocimetry (LDV) measurements. d Acoustic transmission efficiency characterized by equivalent circuit model (EQC) simulations. e Acoustic transmission efficiency characterized by hydrophone experiments. f Comparison of the 2D pressure field between the equivalent circuit model (EQC) and finite element method (FEM) simulations

Article Snippet: Fig. 4 The vibration velocity and two-dimensional pressure field distribution of the single element. a – c Normalized vibration velocities of all cells within one single element characterized by equivalent circuit model (EQC) simulations, COMSOL simulations and laser Doppler velocimetry (LDV) measurements. a Equivalent circuit model (EQC) simulation results. b Finite element method (FEM) simulation results. c Laser Doppler velocimetry (LDV) measurements. d Acoustic transmission efficiency characterized by equivalent circuit model (EQC) simulations. e Acoustic transmission efficiency characterized by hydrophone experiments. f Comparison of the 2D pressure field between the equivalent circuit model (EQC) and finite element method (FEM) simulations In acoustic output characterization of one single element, the axial pressures at 5 mm from the pMUT surface evaluated by the EQC model (Fig. d_i and D_ii) are consistent with the reference data acquired by hydrophone experiments (Fig. e_i and e_ii).

Techniques: Transmission Assay, Comparison

Journal: Microsystems & Nanoengineering

Article Title: A low-voltage-driven MEMS ultrasonic phased-array transducer for fast 3D volumetric imaging

doi: 10.1038/s41378-024-00755-9

Figure Lengend Snippet: Results of the transmission and receiving experiments of a single element

Article Snippet: Fig. 4 The vibration velocity and two-dimensional pressure field distribution of the single element. a – c Normalized vibration velocities of all cells within one single element characterized by equivalent circuit model (EQC) simulations, COMSOL simulations and laser Doppler velocimetry (LDV) measurements. a Equivalent circuit model (EQC) simulation results. b Finite element method (FEM) simulation results. c Laser Doppler velocimetry (LDV) measurements. d Acoustic transmission efficiency characterized by equivalent circuit model (EQC) simulations. e Acoustic transmission efficiency characterized by hydrophone experiments. f Comparison of the 2D pressure field between the equivalent circuit model (EQC) and finite element method (FEM) simulations In acoustic output characterization of one single element, the axial pressures at 5 mm from the pMUT surface evaluated by the EQC model (Fig. d_i and D_ii) are consistent with the reference data acquired by hydrophone experiments (Fig. e_i and e_ii).

Techniques: Transmission Assay

Journal: Microsystems & Nanoengineering

Article Title: A low-voltage-driven MEMS ultrasonic phased-array transducer for fast 3D volumetric imaging

doi: 10.1038/s41378-024-00755-9

Figure Lengend Snippet: a , b Acoustic coupling effects from central excitation and edge excitation characterized by the equivalent circuit model (EQC) model and the FEM model. a Cross-talk analysis of the 3 × 3 array. ( i ) An array using central excitation at element D4, in which neighboring elements B1, B3, and C3 are characterized. ( ii ) An array using edge excitation at element A2, in which neighboring elements A1, A3, and B2 are characterized. ( iii ) Quantitative analysis results for the cross-talk degree (in dB) using the equivalent circuit model (EQC) model and the FEM model. b Cross-talk analysis of the 8 × 8 array. ( i ) An array actuated with central excitation D4, in which neighboring elements C4, E4, and D5 are characterized. ( ii ) An array actuated with edge excitation B1, in which neighboring elements A1, C1, and B2 are characterized. ( iii ). Quantitative analysis results for the cross-talk degree (in dB) using the equivalent circuit model (EQC) model and the FEM model. c – f Varied focusing intensities of the 8 × 8 MEMS phased-array transducer characterized by equivalent circuit model (EQC) simulations and experimental measurements. c Focused pressure at different depths: Comparison between the calculated results of the equivalent circuit (EQC) model and the experimental measurements. d Spatial pressure field distribution of the entire array calculated by the equivalent circuit (EQC) model. e Relationship between the excitation voltage amplitude set in the imaging platform and the focused pressure measured at 30 mm. f Relationship between the actual voltage amplitude on the device and the focused pressure measured at 30 mm

Article Snippet: Fig. 4 The vibration velocity and two-dimensional pressure field distribution of the single element. a – c Normalized vibration velocities of all cells within one single element characterized by equivalent circuit model (EQC) simulations, COMSOL simulations and laser Doppler velocimetry (LDV) measurements. a Equivalent circuit model (EQC) simulation results. b Finite element method (FEM) simulation results. c Laser Doppler velocimetry (LDV) measurements. d Acoustic transmission efficiency characterized by equivalent circuit model (EQC) simulations. e Acoustic transmission efficiency characterized by hydrophone experiments. f Comparison of the 2D pressure field between the equivalent circuit model (EQC) and finite element method (FEM) simulations In acoustic output characterization of one single element, the axial pressures at 5 mm from the pMUT surface evaluated by the EQC model (Fig. d_i and D_ii) are consistent with the reference data acquired by hydrophone experiments (Fig. e_i and e_ii).

Techniques: Comparison, Imaging