Jua.ai’s Post

Machine learning + Materials development 🚀 It was an exciting week at Ljubljana! Autonomous laboratories, material discovery, and manufacturing optimization. It is hard to describe every amazing thing I saw at @Machine Learning Modalities for Materials Science (ML4MS 2024)! 💎 During the conference, I had an amazing opportunity to present my work on the atomic-scale simulations of #alloys. Specifically, my research covers metallic glasses – both challenging and application-promising materials. I am excited that I can leverage the power of digital tools💻to boost materials development 📈! I could not achieve those results without the support of the team from NOMATEN (Silvia Bonfanti, Anshul D. S. Parmar, and Mikko Alava), and Warsaw University of Technology (Jan S. Wróbel). Thanks to Jozef Stefan Institute for hosting such an amazing event, and DAEMON COST for providing me the opportunity to make this trip 🌍. I am looking forward to more such amazing events ✨ #MachineLearning #MaterialsScience #AI #RnD #Research #Innovation #MetallicGlasses

Just had an amazing experience at the 3rd edition of the Summer School on 'The Era of #AI and #Digitalization for Structural Applications'! 🚀 Learned so much about how AI is transforming structural engineering and met some incredible people. #AI #Digitalization #Engineering #Innovation #StructuralHealthMonitoring #Composites

To celebrate National Science Week, we asked the question, “What does science mean to you?”: “To me, even the word 'science' has always had a cool/interesting connotation. I grew up reading Popular Science, New Scientist, and Popular Mechanics magazines. Science was my favorite subject at school, and I usually ended up doing chemistry experiments on my bedroom floor which left burn marks and stains on the carpet (much to the dismay of my parents, but they still encouraged my curiosity). This set me on a path towards a PhD in Robotics (a Computer Science degree), a Research Scientist career at CSIRO, and ultimately, the co-founding of Emesent. My relationship with science has evolved through those roles, but overall I've seen science as a process of taking existing understood, proven knowledge as a starting point and building on that by testing hypotheses to learn something new. This forms the starting point for the next scientist to build on. I really enjoy this process of building, learning, and making sense of things.” - Co-Founder and CSO, Stefan Hrabar, Emesent - #Emesent #Hovermap #ScienceWeek #LiDAR #SLAM

Yes, we are part of the Flanders AI Research (#FAIR) initiative! Yesterday was the yearly Research Day happening at Wintercircus Ghent. Prof. Guillaume Crevecoeur with his team contributes by developing novel methods on #hybrid #modelling (combining physics based models with data driven approaches) and #hybrid #control (combining conventional control methods with RL). Tom Lefebvre gave a deep dive on the latter topic. Contributions from Cedric Van Heck and Victor Vantilborgh. Prof. Jeroen De Kooning and Kurt Stockman with their team are creating useful datasets for the research (thank you Jasper De Viaene) and are using the demonstrator to implement our routines at a higher TRL level (thank you Yentl Thielemans for the slides). This work is a.o. done in the Flanders Make projects #CADAIVISION, #REKPEK, #QUASIMO. The demonstrator is a 2x2x1m³ winding machine processing multiple materials such as paper, plastic foil, ... Vlaams AI Onderzoeksprogramma / Flanders AI Research Program Ghent University Faculty of Engineering and Architecture UGent Flanders Make

John Hopfield and Geoffrey Hinton won the Nobel Prize in Physics. They received the prize for machine learning models based on physics: the Hopfield network and the Boltzmann machine. Both are based on physical models for learning. Yes, some people argued that their work is not a direct physical contribution. However, physics has always been foundational and present since the beginning of artificial intelligence. An example for this; is the Monte Carlo Simulation, which remains a strong model to this day, along with many other examples. In general, I would like to mention specific point that came to my mind earlier, which is applying machine learning models to #education. However, the problem I encountered was identifying the features we could use to measure a certain variable. For example, when measuring a student's performance, we face the challenge of selecting features, especially those related to the psychological aspect. Here, we must rely on psychology-based references. I mean there Should at least a research team that agree on certain features to assess student #behaviors, with a specific weight assigned to each feature. Moreover, when measuring a student's performance, we could at least aim to be closer to reality. This issue is what has halted my progress, despite starting to think about this project six months ago. #ArtificialIntelligence #Physics #education #School #Psychology

I'm excited to share that I'll be presenting my work at the Computational Mechanics / 4th Pan American Congress on Computational Mechanics (WCCM-PANACM 2024) conference! During my presentation, I'll be discussing: - The application of physics-informed neural networks (PINNs) for metal additive manufacturing - Techniques to enhance temperature prediction accuracy and process adaptability - The integration of real-time data for improved thermal modeling and control in AM processes Title: "Adaptive online learning with physics-informed neural networks for thermal prediction in metal additive manufacturing" Authors: Pouyan Sajadi, Mostafa Rahmani Dehaghani, Yifan Tang, and Prof. Gary Wang Date & Time: July 24, 10:25 - 10:45 AM Session: 1825 Room: 213 If you're attending the conference, I would love for you to join my session. Alternatively, let's catch up over a coffee to discuss this research. Looking forward! #ComputationalMechanics #WCCM2024 #PANACM2024 #MetalAM #AdditiveManufacturing #PhysicsInformed #PINNs

The high dimensional encoding, the ability to solve mixed integer and continuous variable problems, and all-to-all connectivity make Dirac-3 a unique and promising optimization machine.

🌟 Exciting News from the World of Physics! 🌟 This year’s Nobel Prize in Physics has been awarded to John Hopfield and Geoffrey Hinton for their groundbreaking contributions that laid the foundation for modern machine learning. - John Hopfield developed the Hopfield network, a model that enables the storage and reconstruction of patterns, akin to human associative memory. This network can recognize incomplete or noisy data, making it a powerful tool for data analysis. - Geoffrey Hinton introduced the Boltzmann machine, which learns from examples rather than explicit instructions. By leveraging principles from statistical physics, this model identifies patterns and probabilities within datasets. 🧠 Machine Learning vs. Traditional Software: While traditional software processes data through predetermined steps, machine learning allows computers to learn from examples, tackling complex problems like image recognition and language translation. 🔬 Impact on Physics and Beyond: The methods pioneered by Hopfield and Hinton are revolutionizing fields such as physics, aiding in the analysis of vast datasets for discoveries like the Higgs particle and gravitational waves. Their work is also paving the way for advancements in molecular predictions and material efficiency. 🌐 Looking Ahead: As we continue to harness the power of machine learning, ethical considerations will be paramount. Responsible use of these technologies is crucial for their sustainable development. Congratulations to this year’s laureates for their monumental contributions! 🎉 #NobelPrize #Physics #MachineLearning #ArtificialIntelligence #Innovation #DataScience