Dear Jousef Murad, I hope this email finds you well. I wanted to take a moment to express my heartfelt gratitude for the incredible podcast series you've been providing, offering year-long experiences and insights from the expert cohort. It's truly been a gift to have access to such valuable content for free. AS AN INDIAN SUBSCRIBER to your RUclips channel, I can confidently say that Indian students, myself included, deeply appreciate the eye-opening industry-driven content you create. Your dedication to sharing knowledge and experiences is commendable and incredibly valuable to aspiring professionals worldwide. Thank you once again for your hard work and generosity in sharing your journey and expertise with us. Looking forward to more enlightening content from you in the future!
Very well put dicussion. PINNs are not ready yet for industry problems, for the Forward cases at least. The convergence is not guaranteed and then comes the additional burden of hyperparaneter tuning and architecture optimization. We should rather focus its development on exploiting it where conventional solvers fall weak, ie; inverse problems. Integrating Inverse PINNs to conventional numerical solvers in order to accelerate calibration of the model seems the way forward. Thanks for the nice episode Jousef.
![Physics Informed Neural Networks (PINNs) [Physics Informed Machine Learning]](/img/1.gif)
📸 IG: instagram.com/jousefmrd/
🎬 Full tutorial on PINNs: ruclips.net/video/G_hIppUWcsc/видео.html
I really liked the way the guest talked about use cases for PINNS Vs traditional methods. Very on point.
Dear Jousef Murad,
I hope this email finds you well. I wanted to take a moment to express my heartfelt gratitude for the incredible podcast series you've been providing, offering year-long experiences and insights from the expert cohort. It's truly been a gift to have access to such valuable content for free.
AS AN INDIAN SUBSCRIBER to your RUclips channel, I can confidently say that Indian students, myself included, deeply appreciate the eye-opening industry-driven content you create. Your dedication to sharing knowledge and experiences is commendable and incredibly valuable to aspiring professionals worldwide.
Thank you once again for your hard work and generosity in sharing your journey and expertise with us. Looking forward to more enlightening content from you in the future!
Appreciate the kind words my friend! 🙂
Very well put dicussion. PINNs are not ready yet for industry problems, for the Forward cases at least. The convergence is not guaranteed and then comes the additional burden of hyperparaneter tuning and architecture optimization. We should rather focus its development on exploiting it where conventional solvers fall weak, ie; inverse problems. Integrating Inverse PINNs to conventional numerical solvers in order to accelerate calibration of the model seems the way forward.
Thanks for the nice episode Jousef.
👍🏻
'Promosm' 😏