Andrew O'Shei

Yann LeCun's vision for autonomous intelligence revolves around a modular, configurable architecture. This highlights the importance of self-supervised learning for AI to learn world models without needing extensive labeled data. Here’s a detailed look at its components: ▪️ Configurator: At the heart of this system is the Configurator, which acts as the AI's executive control center. It dynamically adjusts other components based on the task or context, optimizing the AI's performance. ▪️ Perception module: The Perception module captures and analyzes sensory data from multiple sensors, estimating the current state of the world. This is foundational for higher-level processing and decision-making. ▪️ World model module: It predicts future states of the environment and fills in missing information. Acting as a simulator, it uses current and past data to forecast future conditions and possible scenarios. This enables hypothetical reasoning, planning and more. ▪️ Cost module: Evaluates potential consequences of actions. It has two submodules: - Intrinsic cost: Hard-wired, calculating immediate risk - Critic: Trainable, estimating future costs based on current actions ▪️ Actor module: The Actor module proposes actions based on predictions and evaluations from other components. It computes optimal sequences to minimize costs, using methods similar to optimal control theory. ▪️ Short-term memory: Short-term memory tracks the immediate history of the system’s interactions with the environment. It stores recent data on the world state, actions taken, and associated costs. This allows real-time decision-making. Read our article from the AI 101 series to explore how this concept helps form JEPA: https://lnkd.in/ehqXW6T7

1

🌟 𝗘𝗺𝗽𝗼𝘄𝗲𝗿 𝗬𝗼𝘂𝗿 𝗗𝗶𝗴𝗶𝘁𝗮𝗹 𝗧𝘄𝗶𝗻 𝗣𝗿𝗼𝗷𝗲𝗰𝘁𝘀 𝘄𝗶𝘁𝗵 𝗟𝗮𝗿𝗸𝗫𝗥! 🌟 Are you a 3D XR #Developer working on #DigitalTwin projects? Discover how LarkXR can transform your work with unparalleled precision and collaboration capabilities. #3DXR 𝗪𝗵𝘆 𝗟𝗮𝗿𝗸𝗫𝗥? 🎯 𝗣𝗿𝗲𝗰𝗶𝘀𝗶𝗼𝗻-𝗘𝗻𝗴𝗶𝗻𝗲𝗲𝗿𝗲𝗱 𝗩𝗶𝘀𝘂𝗮𝗹𝗶𝘇𝗮𝘁𝗶𝗼𝗻𝘀 - Achieve accurate virtual representations of physical assets. Leverage advanced analytics and simulations for high-precision results. 💡 𝗗𝗲𝘃𝗲𝗹𝗼𝗽𝗲𝗿 𝗘𝗺𝗽𝗼𝘄𝗲𝗿𝗺𝗲𝗻𝘁 - Utilize developer-friendly tools and a rich API/SDK for seamless integration. Benefit from zero-code deployment and flexible cloud infrastructure options. 🤝 𝗖𝗼𝗹𝗹𝗮𝗯𝗼𝗿𝗮𝘁𝗶𝘃𝗲 𝗜𝗻𝘁𝗲𝗹𝗹𝗶𝗴𝗲𝗻𝗰𝗲 - Enable seamless multi-user interactions in collaborative control room scenes. Support unified operations and manage high concurrency scenarios effectively. 🌐 𝗟𝗶𝗴𝗵𝘁𝘄𝗲𝗶𝗴𝗵𝘁 𝗘𝘅𝗽𝗲𝗿𝗶𝗲𝗻𝗰𝗲 & 𝗗𝗶𝘃𝗲𝗿𝘀𝗲 𝗗𝗶𝘀𝘀𝗲𝗺𝗶𝗻𝗮𝘁𝗶𝗼𝗻 - Enjoy adaptability and flexibility with no terminal restrictions. Deliver refined experiences across any device, including #WebXR, SR/VR/AR/MR, and #spatial devices. 🚀 𝗘𝗻𝘁𝗲𝗿𝗽𝗿𝗶𝘀𝗲-𝗥𝗲𝗮𝗱𝘆 𝟯𝗗 𝗫𝗥 𝗥𝗲𝗻𝗱𝗲𝗿𝗶𝗻𝗴 𝗦𝗼𝗹𝘂𝘁𝗶𝗼𝗻𝘀 - Optimize performance with multi-#GPU resource allocation and fine-grained scheduling. Achieve ultra-low latency (<30ms) with adaptive bitrate adjustments for flawless performance. 🎨 𝗛𝗶𝗴𝗵-𝗙𝗶𝗱𝗲𝗹𝗶𝘁𝘆 𝗮𝗻𝗱 𝗛𝗶𝗴𝗵-𝗣𝗿𝗲𝗰𝗶𝘀𝗶𝗼𝗻 𝗖𝗼𝗻𝘁𝗲𝗻𝘁 - Render intricate #3Dcontent directly in the browser supporting #WebGL application while ensuring unmatched visual fidelity. Support massive scalability with 20,000+ concurrent sessions and 50+ million minutes of stable operation. ⚙️ 𝗦𝘁𝗿𝗲𝗮𝗺𝗹𝗶𝗻𝗲𝗱 𝗣𝗿𝗼𝗷𝗲𝗰𝘁 𝗗𝗲𝗽𝗹𝗼𝘆𝗺𝗲𝗻𝘁 & 𝗥𝗲𝘀𝗼𝘂𝗿𝗰𝗲 𝗘𝗳𝗳𝗶𝗰𝗶𝗲𝗻𝗰𝘆 - Optimize #rendering resources for complex designs to speed up iterations and refinements. Flexible delivery models for cost-effective scaling from prototyping to deployment. 𝗜𝗻𝗱𝘂𝘀𝘁𝗿𝗶𝗲𝘀 𝗕𝗲𝗻𝗲𝗳𝗶𝘁𝗶𝗻𝗴 𝗳𝗿𝗼𝗺 #LarkXR 🌍 🏗️ #AEC (Architecture, Engineering, Construction) 🚗 #Manufacturing, Automotive, Supply Chain 🏥 Healthcare, Education, Retail ⚡ Utility (Mining, Maritime, Energy, Petroleum) 🌾 Agriculture, Aerospace, #SmartCities Start Your Journey with #LarkXR 🚀✨ 🌐 ➤ www.paraverse.cc 💬 @ business@paraverse.cc #3DRendering #RealTimeRendering #XR #3DVisualization #XRDeveloperTools

4

🔴 [LIVE] Join us live at NVIDIA Developer Day with Scaleway this morning! Fabien Da Silva and Javier Figueroa shared invaluable tips on accelerating your workloads with the right NVIDIA GPUs available at #Scaleway! One of them was, “To train a foundational LLM, you need money, lots of data, a skilled team, expertise, and plenty of compute power.” - Fabien Da Silva. Adding the fact that “Scaleway GPU Instances Work like a charm!” according to Javier! 💪 Don’t miss out on several sessions this morning. Tune in now!   #Webinar #AI #GPU

22

3 commentaires

#NVIDIAOmniverse allows developers to integrate #OpenUSD into existing tools, fundamentally transforming their #3D workflows. Watch the latest foundational tech montage from #COMPUTEX24. 👀 https://nvda.ws/3RdNSfp

312

4 commentaires

Bingbot To Test Zstd Compression After Fully Gaining Full Brotli Compression Fabrice Canel from Microsoft announced that BingBot now fully supports Brotli compression and will soon be testing zstd Zstandard compression, a lossless data compression, for its crawler. https://lnkd.in/d6m3iWag #PPC #digitalmarketing #google #marketing #socialmediamarketing #socialmedia #business #seo #branding #marketingdigital #onlinemarketing #entrepreneur #instagram #advertising #contentmarketing #marketingstrategy #digitalmarketingagency #digital #marketingtips #smallbusiness #digitalmarketingtips #website #marketingagency #startup #motivation #success #ecommerce #onlinebusiness #ads

1

𝗘𝗻𝗵𝗮𝗻𝗰𝗲𝗱 𝗹𝗶𝗴𝗵𝘁𝗶𝗻𝗴 𝘀𝗲𝘁𝘁𝗶𝗻𝗴𝘀 𝗶𝗻 𝗧𝘄𝗶𝗸𝗕𝗼𝘁 𝟱 We've enhanced how light sources are set within the graph editor of TwikBot 5, allowing users to easily add and adjust them through nodes. This improvement accelerates the setup of realistic scenes for your 3D product configurators, ensuring they are more visually compelling than ever. Here’s an overview of the enhanced lighting options available:   💡 𝗔𝗺𝗯𝗶𝗲𝗻𝘁 𝗟𝗶𝗴𝗵𝘁 Provides uniform illumination across all objects in the scene, preventing any area from being in complete darkness. 🌈 𝗛𝗲𝗺𝗶𝘀𝗽𝗵𝗲𝗿𝗲 𝗟𝗶𝗴𝗵𝘁 Simulates natural outdoor light with a gradient effect from a sky color to a ground color. This type of lighting is particularly effective for outdoor scenes or simulating daylight indoors. ☀️ 𝗗𝗶𝗿𝗲𝗰𝘁𝗶𝗼𝗻𝗮𝗹 𝗟𝗶𝗴𝗵𝘁 Mimics sunlight with parallel rays coming from a specific direction. Ideal for casting strong, defined shadows and highlighting textures. 🎯 𝗦𝗽𝗼𝘁𝗹𝗶𝗴𝗵𝘁 Emits a focused cone of light, similar to a stage spotlight, perfect for emphasizing specific areas or objects. 🔦 𝗣𝗼𝗶𝗻𝘁 𝗟𝗶𝗴𝗵𝘁 Radiates light in all directions from a single point, similar to a bare bulb. Versatile for simulating various indoor lighting scenarios.   TwikBot 5 allows users to adjust parameters such as color, intensity, and position of each light source, ensuring perfect lighting effects for any product scene.   Elevate your 3D product visualization with TwikBot 5’s advanced lighting settings, and create immersive, stunning 3D product configurator experiences. 🚀 #Twikit #TwikBot5 #3DProductConfigurators #3DProductVisualization

15

1 commentaire

Join us tomorrow for Lecture 21 in Digital Design and Computer Architecture. We'll look at how to exploit data parallelism in SIMD and GPU Architectures. We'll discuss Flynn's taxonomy of computers, array vs. vector processors, SIMD extensions in modern systems, SIMD operations in modern (Machine Learning) accelerators and look at many design examples. Everyone is welcome to join us online! Livestream Thursday May 16 2:15PM Zurich time (CEST): https://lnkd.in/ejxjh6Ei DDCA Spring 2024 website: https://lnkd.in/eCbewkX8 Mohammad Sadrosadati, Frank Kagan Gurkaynak, PULP Platform, Onur Mutlu ETH Zurich, Department of Information Technology and Electrical Engineering #education #computing #computerscience #engineering #computers #hardware #software #design #paradigms #innovation #research #computerarchitecture #memory #storage #communication #systems #architecture #machinelearning #FPGAs #GPUs #security #reliability #robustness #data

19

⚡️Multi-Agent RAG Online Workshop⚡️ If there’s anything better than agentic RAG, it’s multi-agent RAG! In this event, we’ll explore the big idea behind “multi-agent” applications. These types of workflows combine multiple independent agents, which can be structured to work together to solve complex problems. The patterns of using multi-agent frameworks can take many forms including planning, reflection, and tool use. LangGraph, our library for building stateful, multi-actor applications with LLMs, can help you orchestrate your system. Sign up here ➡️ https://lu.ma/agentrag

236

3 commentaires

📃Scientific paper: A Practical Guide to Implementing Off-Axis Stereo Projection Using Existing Ray Tracing Libraries Abstract: Virtual reality (VR) renderers driving CAVEs and similar immersive environments use the off-axis stereo camera model so that a tracked user can move freely in front of the projection plane. Geometrically, off-axis projection results in asymmetric viewing frusta and generalizes the ubiquitous perspective camera model to support positioning off the center of the projection plane. VR renderers often integrate with larger visualization systems that rely on libraries for position tracking and pose estimates, for ray tracing-based rendering, and for user interaction. We demonstrate different strategies to implement off-axis stereo projection within the constraints of given VR applications and ray tracing libraries. We aim for minimal to no adjustments required to the internal camera representation of such libraries. We include host and shader code with the article that can be directly integrated in custom applications. Continued on ES/IODE ➡️ https://etcse.fr/xv9 ------- If you find this interesting, feel free to follow, comment and share. We need your help to enhance our visibility, so that our platform continues to serve you.

1

Here is a very useful systematic survey on prompting: https://lnkd.in/gp-ZzGrD I take this opportunity to  point at our work deploying prompting for soundscape generation back in 2012. References as well as access to an archived working version can be found here: https://lnkd.in/gTuEqyUN Unsurprisingly, it is missing from the Section on the history of prompting. It is important to note that prompting is not tied to language models, nor it is to neural networks per se. The survey is a very useful read nonetheless. For those less familiar with the topic, I guaranty that you will get ideas of things to try in your domain reading it. I am curious to know what you will get out of it. #genAI #AI #prompting

12

1 commentaire

Computer vision is reshaping the landscape of 3D virtual production in the realm of video production. Here's why producers can't ignore #computervision and #AI. 1/ Cutting-edge algorithms now can reconstruct precise 3D models from mere 2D images utilizing methods such as photogrammetry and neural radiance fields (NeRFs). This facilitates the rapid generation of digital assets and environments with accuracy. 2/ An important use case lies in markerless motion capture, where computer vision allows to track an actor's movements solely through camera inputs. This process seamlessly translates the performance onto a digital avatar in real time, eliminating the need for cumbersome motion capture suits. 3/ In the realm of LED virtual production volumes, computer vision plays a pivotal role in enabling the seamless integration of physical elements with 3D graphics rendered by game engines in real time. Through meticulous camera tracking, the virtual camera aligns with the physical rig, creating a seamless fusion of reality and CGI. 4/ The integration of generative AI models with computer vision technology now facilitates the swift creation of photorealistic 3D environments and objects for virtual productions, significantly reducing both lead times and costs compared to conventional 3D asset creation pipelines. 5/ Utilizing methods such as laser scanning and photogrammetry, computer vision empowers filmmakers to effortlessly incorporate realistic digital representations of real-world objects into virtual sets and scenes. 6/ Furthermore, with the aid of AI-driven tools, labor-intensive processes like rotoscoping and marker/object tracking for compositing are now automated through computer vision, streamlining VFX workflows. 7/ As computer vision algorithms continue to advance, we anticipate virtual productions achieving heightened levels of realism and fidelity that were once unattainable or excessively costly through traditional VFX techniques. 8/ What was once exclusive to high-budget Hollywood productions is now within reach of indie filmmakers and content creators. Computer vision is democratizing 3D content creation and virtual production in an unprecedented manner - marking a truly transformative advancement in this field.

2

1 commentaire

Your FAQs About RoT STUDIO Answered! Are you curious about RoT STUDIO and how it can revolutionize your virtual reality and extended reality projects? We've got you covered! Swipe through our carousel to get answers to the most frequently asked questions from our customers. 👉 Do I need to know coding to use RoT STUDIO? 👉 Can I do modeling in RoT STUDIO? 👉 How do I create and find ready-made patterns? 👉 Does HDRI work in RoT STUDIO? 👉 What is the minimum computer configuration needed? 👉 How can we run trainings designed with RoT STUDIO? 👉 What should we pay attention to when choosing VR hardware? Find out all these answers in our post! What feature of RoT STUDIO are you most excited to try out? Comment your thoughts down below! 💬 Reach out to us now for new possibilities for growth and innovation for your organization with our VR and XR solutions.  🌐 https://lnkd.in/gKCmFYW4 ☎️ +90 312 4000 768 ✉️ info@rotstudio.com #virtualreality #vr #extendedreality #xr #FAQs #technology #innovation

15

📣 Exciting News: Atelier B Community Edition 24.04 Available! 😀 We are thrilled to announce the release of our latest version of Atelier B. Here's what's new and improved in this version: 💡Added Functionalities 🔑 Automatic Proof-Support: Utilize off-the-shelf automatic theorem provers (Alt-Ergo, CVC4, CVC5, iProver, Vampire, Z3) for proof obligations. 🔑Generate Rust code from software projects. ➡ For detailed release notes and downloading the software, visit https://lnkd.in/dWdhxfHB. ➡Windows version is available now. Versions for several Linux distributions are coming soon, followed by a macOS version. 😉 Thank you for your continued support and feedback! ➡Download the latest version now and share your experience with us. #formal #programming #formalmethods #safetycritical #development #mathematicalproof #softwaredevelopment #formallanguages #proof #CLEARSY

33

2 commentaires

You asked for it, so we built it: ✏️ Edit Annotations This update allows you to adjust pins while creating a new annotation, or edit existing ones on 3D models, without starting from scratch. Your annotation measurements will automatically be updated, and history will show it has been updated. Watch a speed demo below. 👉 Read more in our July product update: https://lnkd.in/gXHzmv7q

36

1 commentaire

The world of 3D modelling is rapidly evolving with the advent of NVIDIA Instant NeRFs. This technological advancement revolutionizes the way we perceive, create, and interact with 3D models. Shake up your day with a dose of insight https://lnkd.in/g_drpjWR #BrandXR #3D #3Dmodeling #3Ddesign #NVIDIA #NeRFs

1

Topology and adaptive Level-of-Detail (LOD) have been unsolved problems and critical roadblocks towards full-scale studio/enterprise adoption of 3D GenAI. A new era in 3D generative modeling is taking shape, beyond SDS-based NeRFs/Gaussian Splats, SDFs, DMTeT, Flexicubes, etc. Here is an early sneak peak. We want to go deep with a small group of partners on this. You know what to do: hello@csm.ai

308

3 commentaires

Big changes to " unity " XRI 3.x at hand ! To move towards more modular data + input better aligned with current " open xr " hand actions. Its also a work around for " vision pro " not providing physics raycasts through os inputs. Plus also consolidates near and far interactions into " attach transforms ".. The prior " affordances " system is now deprecated and will be removed in XRI 5.x as we shift from controller - based to pure " open xr " hand interactions. That also solves the problem of input hierarchies. My friend Eric from " unity " will do a proper live stream later this month. For now the gdc talk has just gone up and details the new architecture .. REF : > GDC 2024 " unity XRI 3.0 " session [ https://lnkd.in/gKAaaV63 ]

4