Model-Based and Physics-Informed Deep Learning Neural Network Structures
www.mdpi.com/2673-9984/12...
By Ali Mohammad-Djafari et al.
From the MaxEnt 2024 Workshop
#NeuralNetworks #DeepLearning #MachineLearning #PINNs
Researchers from Queensland University of Technology, Tsinghua University, and international partner institutions reported their findings in Acta Mechanica Sinica.
#PINNs #AI
Details: doi.org/10.1007/s104...
125 citations: Can you trust your PINN solutions? This work develops rigorous error estimates for residual minimization in neural networks, providing convergence guarantees for quantifying solution accuracy.
📖 www.dl.begellhouse.com/journals/558...
#ScientificML #PINNs
We're growing my group and #hiring! Up to four positions:
- #LLM agents for scientific discovery processes
- #uncertainty modeling, #PINNs, adversarial robustness
- #foundation models for #astrophysics and #particle physics
- #ML for #astro
Write lflek@uni-bonn.de or lt's talk at #EurIPS
We are growing my group and #hiring! Up to four positions:
- #LLM agents for scientific discovery processes,
- #uncertainty modeling, #PINNs, adversarial robustness,
- #foundation models for #astrophysics and #particle physics,
- #ML for #astro
Physics and ML background welcome.
lflek@uni-bonn.de
The past two day we attended the annual symposium organised by the @barcelonacollaboratorium.com
Our PhD student, Júlia Vicens-Figueres, has presented a flash talk and a poster about modeling bacterial response to antibiotics with physics-informed neural networks #PINNs
#Causality#AI#Biology
🚀 Aleix Fornieles (Eurecat) shared how Physics-Informed Neural Networks boost hydrogen yield from biomass gasification—merging physics + AI for cleaner energy!
#AI #Hydrogen #Sustainability #PINNs #CleanEnergy
RBF-PIELM Shows Speed Gains Over PINNs for Biharmonic Equation
RBF‑PIELM trained 350× faster than standard PINNs and used over 10× fewer parameters for the biharmonic equation. The work was accepted at NeurIPS ML & Physical Sciences Workshop. Read more: getnews.me/rbf-pielm-shows-speed-ga... #rbfpielm #pinns #neurips
Physics Informed Neural Networks Linked Directly to External Solvers
A new method lets Physics Informed Neural Networks incorporate exact residuals from external forward solvers as a loss term, removing a key obstacle. The work appeared in September 2025. getnews.me/physics-informed-neural-... #pinns #physics
Assessing PINNs' Ability to Handle Noise in Inverse Fluid Mechanics Problems
A study finds physics‑informed neural networks need less setup than a finite‑element‑optimizer, but FEM yields higher accuracy and faster compute times on fluid‑mechanics inverse. Read more: getnews.me/assessing-pinns-ability-... #pinns #fem
Thermodynamically Informed Neural Networks Boost Physics‑Informed AI
THINNs (Thermodynamically Informed Neural Networks) use large‑deviation theory to weight penalties, not heuristic loss. Benchmarks show lower residual errors. Read more: getnews.me/thermodynamically-inform... #thermodynamics #neuralnetworks #pinns
Error Estimates for PINNs Solving the Boltzmann Equation
New research gives the first rigorous error bounds for PINNs solving the Boltzmann equation near a global Maxwellian and shows the method keeps the asymptotic-preserving property. getnews.me/error-estimates-for-pinn... #pinns #boltzmann
Second‑Order Optimization Aligns Gradients in PINNs
A new SOAP optimizer using second‑order preconditioning delivers 2–10× accuracy gains on PINNs and handles turbulent flows up to Reynolds 10,000, achieving state‑of‑the‑art results on ten PDE benchmarks. getnews.me/second-order-optimizatio... #pinns #soapoptimizer
Trainable Activation Functions Boost PINN Loss Balancing in Fluid Flow
Trainable activation functions with adaptive loss weighting cut RMSE errors by 7.4%–95.2% on Navier‑Stokes fluid‑flow tests, per a preprint released 17 September 2025. Read more: getnews.me/trainable-activation-fun... #pinns #machinelearning
PBPK-iPINNs: Physics-Informed Neural Networks for Brain Modeling
Researchers introduced PBPK‑iPINN, combining physics‑informed neural networks with PBPK models to infer drug parameters from concentration points, achieving accuracy comparable to solvers. Read more: getnews.me/pbpk-ipinns-physics-info... #pbpk #pinns
Boundary Element Method vs PINNs for Wave Scattering
A study comparing BEM and PINNs finds PINNs need about 42× longer training than BEM but evaluate up to 204× faster, while BEM keeps stable error across larger domains. Read more: getnews.me/boundary-element-method-... #wavescattering #bem #pinns
AI‑Driven Model Predictive Control for SIR Epidemic Management
A new framework fuses PINNs with MPC to estimate SIR epidemic states from noisy data. Scenario A assumes a known recovery rate; Scenario B uses a known R₀. Read more: getnews.me/ai-driven-model-predicti... #pinns #mpc
Check out my newest article on Towards Data Science!!!
@towardsdatascience.com #PINNs #Physics
towardsdatascience.com/physics-info...
7/7 Read the preprint here: arxiv.org/abs/2507.08972
#PINNs #CFD #Turbulence #ScientificComputing #MachineLearning #DOE #ASCR #AppliedMathematics #Yale #UPenn #PNNL
It would be great to be able to see a compiles list of useful PDEs that #PINNs struggle to solve - and how would we measure success there.
We know of edge-cases with simple PDEs, where PINNs struggle, but then often those aren't the cutting-edge of use-cases of PDEs.
Adaptive Physics-informed Neural Networks: A Survey
Edgar Torres, Mathias Niepert
Action editor: Stratis Gavves
https://openreview.net/forum?id=vz5P1Kbt6t
#adaptive #pinns #pdes
Learn from the MATLAB experts how to implement these techniques in your next project. See you there! #PhysicsAI #MATLAB #MachineLearning #PINNs
1. Physics-Informed Neural Networks (PINNs): Teaching AI the laws of physics for predictions that respect reality 🧠 #PINNs
2. Fourier Neural Operator (FNO): Transforming complex systems into frequency domains where patterns emerge 🔍 #FourierAI
New #Survey Certification:
Adaptive Physics-informed Neural Networks: A Survey
Edgar Torres, Mathias Niepert
https://openreview.net/forum?id=vz5P1Kbt6t
#adaptive #pinns #pdes
🚀 Introducing #PINNverse — a game-changer for parameter estimation in differential equations! 🧠💡
No forward solves. Better accuracy. Robust to noise.
Preprint: doi.org/10.48550/arX...
#SciComm #MachineLearning #InverseProblems #PINNs
Join the AI for Science Bootcamp (May 27–28, online) to explore Scientific Machine Learning with NVIDIA Modulus!
More details can be found here: www.hlrs.de/training/202...
📅 Register by April 18: gpuhackathons.org
#SciML #AI #HPC #PINNs #NVIDIA #Modulus #EuroCC #HLRS
My first year PhD student Sébastien André-sloan presents at a #INFORMS conference in Toronto. Its a joint work with Matthew Colbrook at DAMTP, Cambridge. We prove a first-of-its-kind size requirement on neural nets for solving PDEs in the super-resolution setup - the natural setup for #PINNs.
Two weeks ago Júlia participated in the @cs3conference.bsky.social.
Did you know about #PINNs (Physical-Informed Neural Networks)?
📣 Hello everyone 📣
Here on BlueSky I will share my tutorials in datascience, statistics, and AI.
For a nice start, see my tutorial on solving differential equations using neural networks.
labpresse.com/solving-diff...
#AI #Physics #PyTorch #PINNs #NeuralNetworks #DifferentialEquations #science
Fourier PINNs: From Strong Boundary Conditions to Adaptive Fourier Bases
Madison Cooley, Varun Shankar, Mike Kirby, Shandian Zhe
Action editor: Jeremias Sulam
https://openreview.net/forum?id=KqRnsEMYLx
#fourier #boundary #pinns
