physics-informed neural networks
PulseAugur coverage of physics-informed neural networks — every cluster mentioning physics-informed neural networks across labs, papers, and developer communities, ranked by signal.
18 day(s) with sentiment data
-
New AI framework enhances stroke assessment with uncertainty quantification
Researchers have developed Evidential Perfusion Physics-Informed Neural Networks (EPPINN) to improve the accuracy and reliability of computed tomography perfusion (CTP) imaging for acute ischemic stroke assessment. This…
-
PINNs vulnerable to silent failures from parameter misspecification
A new research paper highlights a critical vulnerability in Physics-Informed Neural Networks (PINNs), demonstrating that these models can be misled by incorrect physics parameters during training. This 'parameter poison…
-
AI models predict welding quality across laser and TIG processes · 5 sources tracked
Researchers have developed advanced deep learning models for predicting weld quality in laser and TIG welding processes. One model utilizes a multi-task spatiotemporal deep neural network to predict penetration depth an…
-
New B-PINN framework enhances uncertainty quantification for material degradation prognostics
Researchers have developed a new Bayesian Physics-Informed Neural Network (B-PINN) framework designed to improve uncertainty quantification in prognostics and health management (PHM). This novel approach jointly models …
-
New HSPINN method enhances physics-informed neural network accuracy
Researchers have developed a new method called Adaptive Hard-Soft Physics-Informed Neural Networks (HSPINN) to improve the training and accuracy of physics-informed neural networks (PINNs). Traditional PINNs struggle wi…
-
New PIBLS framework offers faster, more accurate PDE solutions
Researchers have introduced the Physics-Informed Broad Learning System (PIBLS), a novel framework designed to solve partial differential equations (PDEs) more efficiently than existing methods. Unlike traditional numeri…
-
New two-stage evolutionary strategy optimizes PINNs for better accuracy
Researchers have developed a novel two-stage hyperparameter optimization strategy for Physics-Informed Neural Networks (PINNs) to address their sensitivity to hyperparameters and unstable convergence. This approach util…
-
New ModSync framework improves training stability for physics-informed neural networks
Researchers have developed a new training framework called ModSync to address fragility in physics-informed neural networks (PINNs). PINNs, used for solving partial differential equations (PDEs), can become unstable as …
-
New neural networks accelerate PDE solving with improved accuracy and speed · 4 sources tracked
Researchers are developing advanced neural network architectures to improve the solving of partial differential equations (PDEs). One approach, Adaptive Hard-Soft Physics-Informed Neural Networks (HSPINN), enforces boun…
-
Deep learning framework computes geodesic-like curves on parametric surfaces
Researchers have developed a novel framework utilizing deep learning and Physics-Informed Neural Networks (PINNs) to compute geodesic-like curves on parametric surfaces. This method, building on Chen's 2010 work, offers…
-
Scientific Machine Learning advances fluid dynamics simulation
A recent chapter reviews advancements in Scientific Machine Learning (SciML) for simulating complex fluid flow and transport phenomena. It highlights methods like Dynamic Mode Decomposition and Physics-Informed Neural N…
-
Scientific Machine Learning advances fluid dynamics modeling · 2 sources tracked
This chapter explores advancements in Scientific Machine Learning (SciML) for simulating complex fluid flow and transport phenomena. It details methods like Singular Value Decomposition, Dynamic Mode Decomposition, Phys…
-
New LiL-Q method solves nonlinear PDEs with physics-informed neural networks
Researchers have developed a new numerical method called LiL-Q for solving nonlinear partial differential equations (PDEs) using physics-informed neural networks (PINNs). This method employs Bellman-Kalaba quasilineariz…
-
New MODE architecture enhances physics-informed neural networks
Researchers have introduced Manifold-Orthogonal Dual-spectrum Extrapolation (MODE), a novel micro-architecture for adapting physics-informed neural networks (PINNs). MODE addresses limitations in existing methods like S…
-
Dual-Network PINNs Benchmark Optimal Control Problems
Researchers have developed a dual-network Physics-Informed Neural Network (PINN) approach to solve optimal control problems, demonstrated on a mass-spring-damper system. This method achieves accuracy comparable to class…
-
RepNet tackles spectral bias in deep neural networks
Researchers have introduced RepNet, a novel deep neural network architecture designed to address spectral bias, a common limitation in capturing high-frequency and oscillatory behaviors. By reparameterizing the weights …
-
New Koopman-PINN framework enhances epidemic modeling and forecasting
Researchers have developed a new framework called Koopman-PINN that combines Koopman operator theory with physics-informed neural networks for improved epidemic modeling. This approach maps epidemic states into a latent…
-
New methods explore gradient-free optimization for neural networks
Researchers are exploring novel methods for optimizing neural networks without relying on traditional gradient-based approaches. One paper introduces a first-order layer for differentiable optimization that avoids compu…
-
New PINN-FEM coupling method offers theoretical framework for complex simulations
Researchers have developed a novel method for coupling Physics-Informed Neural Networks (PINNs) with Finite Element Methods (FEM) by framing the interaction as a Steklov-Poincaré operator. This approach addresses limita…
-
Direct PDE inversion reveals loss landscape pathology
Researchers explored a direct gradient-based method for inverting reaction-diffusion systems, specifically the Gray-Scott model, by backpropagating loss through the PDE itself. They discovered that this direct approach …