DeepWave@KAUST

🌟 We’re thrilled to announce our successful participation with the DeepWave Consortium booth in IMAGE 2024! 🚀 A great chance to showcase our collective innovation and latest research. A huge thank you to everyone who visited us, and to our sponsors who made this possible. Here’s to many more successful collaborations and innovations! 🌐🤝 #DeepwaveConsortium #Innovation #image24 #Collaboration #Teamwork #Success #kaust

The Consortium's 2024 Annual Meeting concluded and it has been two days filled with insightful interaction with our sponsors! 💡 #DeepwaveConsortium #AnnualMeeting #Houston

Excited to be part of IMAGE 2024 in Houston! Stop by our booth (715) and don't miss our members' presentations and posters. Find detailed list of talks and posters here: https://lnkd.in/d85tB3TT #DeepwaveConsortium #kaust #image24

Excited to be part of IMAGE 2024 in Houston! Stop by our booth (715) and don't miss our members' presentations and posters. Find detailed list of talks and posters here: https://lnkd.in/d85tB3TT #DeepwaveConsortium #kaust #image24

🔬 Exciting breakthrough in seismic reservoir characterization! 🌎 The new paper introduces IntraSeismic, a revolutionary hybrid seismic inversion method that combines implicit neural representations with physics-based seismic modeling. This innovative approach offers: ✅ Unparalleled performance in static and dynamic 2D/3D pre-stack and post-stack inversion ✅ Rapid convergence rates ✅ Seamless integration of hard constraints and uncertainty quantification ✅ Potential for data compression and fast randomized model access IntraSeismic's versatility extends beyond seismic inversion - it can be applied to a wide range of seismic and general geophysical inverse problems, opening up new possibilities in various earth science disciplines. Curious to learn more? Check out at https://lnkd.in/d4fjmnXj #SeismicInversion #MachineLearning #Geophysics #EnergyExploration #InnovativeTechnology #DeepwaveConsortium

🚀 Gearing up for the DeepWave Consortium Annual Meeting 2024! We'll be sharing our latest research💡 with our sponsor companies on Aug. 22-23, 2024 in Houston (TX). Find out more here: https://lnkd.in/dGEfSpRZ

📢 During EAGE 2024 the Deepwave consortium is delighted to share some of the groups developments in promoting machine learned solutions to seismic problems, covering topics ranging from seismic and full waveform inversion to seismic data processing and modeling and even microseismic monitoring. We hope to show the abilities of some famous ML frameworks, like implicit neural representations, diffusion models, GPT, PINNs, and more in solving some of our outstanding seismic problems, and most in an unsupervised manner, which makes them field data friendly. Check the full list of oral presentations and posters here: https://lnkd.in/ekQaRp7S

🆕 A new paper introducing LatentPINN for wavefield modeling is out! Physics-informed neural networks (PINN) promise a mesh-free approach to solving a specific parametric partial differential equation (PDE). The widely available PINN model to date, however, will have to undergo additional training for a new PDE parameter (e.g., velocity model for the wave equation). To overcome the need for further training, LatentPINN introduces latent representation learning as an additional input to the PINN model. With this additional information, our LatentPINN produces instant (no additional training) PDE solutions for different PDE parameters. Through numerical tests and benchmarking against several existing PINN-based algorithms, we show that our framework provides up to three times the speed up and an order of magnitude accuracy improvement. 🚀 The early access article is accessible through https://lnkd.in/dwU3rNaT

The 3rd issue of the DeepWave Consortium newsletter is out! 🚀 We are excited to share with you the latest updates and developments in our field. 📢 Check out the newsletter here: https://lnkd.in/d5tanWdf Stay tuned for more updates from us!

🚀 A new paper introducing DiffusionEFWI is out! DiffusionEFWI introduces a novel adaptive regularization to the multiparameter elastic full waveform inversion (EFWI) using diffusion models. Through numerous numerical experiments using single-component recordings ranging from (marine) salt to (land) arid environments, we show that DiffusionEFWI improves the parameters estimation and data fitting. DiffusionEFWI shares similar computational costs compared to other regularization techniques while significantly providing a better remedy to the cross-talk issue in EFWI. The paper is freely accessible 📎 here: https://lnkd.in/dJbwemu7