Woongjin Thinkbig Co., Ltd’s Post

The CMU LeCAR Lab will present several papers at #RSS2024 this week, covering topics from learning & control & optimization, safe autonomy, to humanoids. Check them out!

Winning streak had to stop one day 😄 📣📣. 2nd place is still quite good! 📣📣 This time it was about extracting all the objects from an image with GPT4V, which does an incredible job btw, despite a difficult image with quite small or partial objects, or a scarf blending with the sky for instance (see demo) and storing all their relationships in a Neo4J graph to try to mimic how humans (at least me) store information all day long. I wish I had 10,000 GPUs to process 100B images. Here's a recorded demo. https://lnkd.in/eS-RMhAC

Is Reality Fake? The MIND-BLOWING Simulation Theory (Short) Are we living in a giant computer program? The simulation theory explores this mind blowing idea! This short dives into the possibility that our reality might be a simulation. Is it crazy, or could it be true? #simulationtheory #whatif #isitreal #scifi

If you are attending Super Computing this week, be sure to add the MLPerf birds-of-a-feather session this Wednesday to your calendar. Tom St. John and other great MLCommons members will be sharing their insights and learnings. https://lnkd.in/ettSfK6J #SC23 #machinelearning #ai #supercomputing

𝗠𝗲𝗲𝘁 𝗦𝗔𝗠𝗣 𝗮𝘁 𝗜𝗖𝗖𝗩 𝟮𝟬𝟮𝟯 💥 Our Lead Machine Learning Nachwa Bakr, PhD is attending, so let's meet! At Samp we have been processing and hosting hundreds of Industrial #3Dlidar scenes on the web. With the vision of building the product for a multitude of Industrial environments, the typical challenges faced by machine-learning are very large lidar scenes, high density of objects and not large enough Radiance fields ect. Some of the critical 𝗖𝗼𝗺𝗽𝘂𝘁𝗲𝗿 𝗩𝗶𝘀𝗶𝗼𝗻 𝘁𝗮𝘀𝗸𝘀 at the core of our tech are : 👷♀️ Object recognition in point cloud 👷♂️ 3D Geometric shape reconstruction 👩💻 Scene graph representation and understanding 👨💻 Rendering of large 3D lidar scenes on the web Nachwa is looking forward to meeting all the fellow geeks in the Vision AI community, so just ping us if you're in Paris. #ICCV2023 #ComputerVision #Industry #energytransition

Did you know? 🤔 The mathematical name for the division sign (÷) is called an obelus! ➗ Dive into fascinating facts and surprising trivia that will spark your curiosity and expand your knowledge! 🌟 🤓 . . For more details visit our website 👉https://lnkd.in/d_VeeVPg . . DM For Any Queries Call us at +91 7678669938 . . #JBMS #LearnSomethingNew #KnowledgeIsPower #educationmatterters #FunLearning #FunFact #MathTrivia #DidYouKnow #Obelus #MathMagic #LearnSomethingNew #TriviaTime

Another short video showing my ESOL learners producing language and using their phones to display it on a smartboard, ready for editing. They began the lesson by speaking in small groups to tell their shopping stories to each other. Once that was done we're ready for correction and editing. Look out for the followup video in which this work is marked by AI. https://lnkd.in/ev3BcQCa #ESOL #ESL #digitallearning #digital #googleslides #googleforeducation #aiadvancements

It’s always a pleasure to see our NHR4CES members in action: Marcel Krüger, member of our CSG Visualization, presented their recent work at the 8th International Workshop on In Situ Visualization. In this talk the CSG Visualization presented InsitUE - a Catalyst2-compatible library that enables hybrid in-situ/in-transit workflows with Unreal Engine focusing on efficiency and portability. With the introduction of InsitUE, they address the lack of in-situ workflows for modern game engines. An issue raised in one of their recent journal submissions (https://lnkd.in/dV3tc7UY) where they discuss and share insights on how game engines such as Unreal Engine can transform the development process for scientific visualizations and report the experiences gathered over several years in the immersive visualization lab at RWTH Aachen University. InsitUE enables users to combine the Unreal Engine with Catalyst2 instrumented applications which are essential for handling large datasets or obtaining early simulation insights. The paper on InsitUE will be published in the Springer Lecture Notes of Computer Science series soon – stay tuned. #NHR4CES #Visualization #HPC #InsitUE #ISC24 #ReinventingHPC Tim Gerrits Jan Frieder Milke Torsten Kuhlen

Many of our first-time learners are nervous to try a simulation for the first time. Some are reluctant to practice difficult conversations or put themselves through challenging scenarios. But the results speak for themselves! Check out the feedback we received recently from Learners after participating in a Mursion simulation. Book a demo to experience a simulation for yourself – the link is in the comments. #digitallearning #corporatetraining #simulation #professionaldevelopment

I had the great pleasure to present our blogpost "Fair Model-Based Reinforcement Learning Comparisons with Explicit and Consistent Update Frequency" at ICLR last week. With Albert Thomas, Giuseppe Paolo, and Balázs Kégl, we argue in favor of explicit update frequency in model-based reinforcement learning benchmarks, fair comparisons with consistent update frequencies between different algorithms, and more ablation studies evaluating its impact on performance. Check it out here 👉🏻 https://lnkd.in/eSqhdM_C #RL #MBRL #ICLR24