Suhib A.

More classic insights from Joe Easton of Bloomberg News on how NVIDIA has lost over $430 billion in 3 days https://lnkd.in/ebAYvzNd Not all #ai companies are the same. This is a list of funded LLM AI companies of 1400 the majority use third party APIs like OpenAI . What happens if they deprecate their API? Who owns the IP? https://lnkd.in/etXpqd5a At Yellow Sub AI I created all the neural networks from scratch coding individual linear regression algorithms and stitching together deep neural network layers. I then trained them on a unique data set. We don’t use other people’s code. No Matlab. No R. Everything built from the ground up in assembler to ensure the fastest inference possible. And unlike overfitted supervised ML algorithms our unsupervised algorithms can be used on many diverse populations from shopping malls to individual stores through to venues and open spaces #peoplecounting #alpha

5

1 Comment

This is also the XPU using scenario In my mind too. “Whisper on NPU, LLMs on iGPU, document embeddings on CPU! Full use of #AIPC through #OpenVINO to implement scenarios like chat with documents and voice guided assistant that generates code. Everything on our new Lunar Lake platform that was just announced.”

9

Keeping your data close to the GPUs is essential, so we offer VAST Data as part of our #GPUasaService AI Platform. Let us help you #StepontheGaaS and move your #AI strategy forward. #Fast #Secure #Private Dominic Ken Michael Bill Traci

10

1 Comment

A game changing tool for the automotive industry? I clearly see a use case for TTS/STT here as most mechanics would be interested in a hands free experience. I can really see this shining in the context of AR though.

1

Have you heard about tiny AI accelerators (smaller than 1cm x 1cm) that can even run on microcontrollers? I’m excited to share that I’ll be attending #NeurIPS2024 to present Nokia Bell Labs' initiative on building collaborative on-device AI systems: “DEX: Data Channel Extension for Efficient CNN Inference on Tiny AI Accelerators.” This work introduces an innovative approach to improving AI model accuracy without increasing latency, leveraging intelligent data channel extensions on tiny AI accelerators. Huge thanks to Taesik Gong and Fahim Kawsar for their invaluable collaboration! This marks the beginning of our vision to enable multi-device, multi-modal, and multi-sensory computing on the body. Stay tuned for more exciting projects in this direction! * When: Wednesday, Dec. 11, 4:30 PM – 7:30 PM * Where: East Exhibit Hall A-C, Booth 1405 * arXiv: https://lnkd.in/emP4_5vg * Code: https://lnkd.in/ecn5yTCB * Broader overview of our department research: https://lnkd.in/enbW6fnW Summary: Tiny AI accelerators enhance TinyML by improving processing power, but their limited memory often reduces input resolution, negatively impacting accuracy. To address this, we introduce DEX—a method that extends input channels by incorporating spatial information through patch-wise sampling and stacking. DEX leverages unused processors and memory for parallel execution, maintaining inference latency while improving accuracy by an average of 3.5 percentage points.

61

4 Comments

Depth Anything V2 has been released for single image depth estimation 🔥🔥 The increase in accuracy over V1 is incredible and even that one was state-of-the-art by far. Another important element of the Depth Anything family is that they are able to run in real-time. You can just take a single image, video or a stream and run it through the model to get the output depth map for each frame. Improvements and key highlights 📊 ✅ More fine-grained details than Depth Anything V1 ✅ More robust than Depth Anything V1 and SD-based models (e.g., Marigold, Geowizard) ✅ More efficient (10x faster) and more lightweight than SD-based models ✅ Impressive fine-tuned performance with our pre-trained models Check out the project page and GitHub repo here: https://lnkd.in/dUB2bPfH 👇 Personal Website https://lnkd.in/e-KaWkeD

270

1 Comment

Hope you had a chance to see our NVIDIA Holoscan for Media solution on the Monks booth, running real-time brand detection with minimal latency. Under the hood: - Fine-tuned YOLO v8 from the NVIDIA NGC Model Catalog using NVIDIA TAO Toolkit - Integrated with NVIDIA DeepStream SDK as part of the AI Pipeline within NVIDIA Holoscan for Media

44

2 Comments

Thinking why Waymo takes another kind of approach. Do you think Waymo will produce its own vehicles? In the context of self-driving cars, Convolutional Neural Networks (CNNs) play a crucial role in perception tasks, particularly for processing visual data from cameras and other sensors. The CNN architecture enables self-driving systems to identify objects, interpret scenes, and make decisions based on visual inputs. CNNs are integral to self-driving cars, enabling them to process and interpret visual data from the environment. Their architecture supports object detection, scene understanding, semantic segmentation, depth estimation, and even end-to-end driving control, all of which are critical for safe and efficient autonomous driving. The real-time performance, accuracy, and ability to generalize from visual inputs make CNNs a cornerstone of modern autonomous vehicle perception systems. #selfdriving

2

Take this course to learn how to build beautiful, custom UI for 3D tools on NVIDIA Omniverse. You’ll get a chance to build your own custom UI for workflows in Omniverse with hands-on exercises. Take the free course > https://nvda.ws/3AtZLs9

1

🔬 New research from @UTSWMedCenter creates a 4D deep learning #AI model for identifying #breastcancer metastasis using only MRI and clinical data, for early detection without invasive biopsies.🩺 🔗 Learn more: https://nvda.ws/3YppB8V #BreastCancerAwarenessMonth

2

1 Comment

We are all trying to keep up with the pace of #AI #innovation - well, here's a free course from Andrew Ng and Vasudev Lal to help you build LLM and Vision #GenAI pipelines. #IamIntel Cool stuff!

12

We have supported ARM for a long while, persistently and continuously optimizing both our Aurora encoder and Aurora enhancer products to boost their performance on ARM. To support ARM has been essential to meet our customers' requests, (1) for those on mobile devices, both Android and iOS, (2) for those using the MacBook M series, and (3) for those cloud customers on AWS Graviton's. These customer demands have driven us to continuously innovate. After deeply optimizing our AI-driven encoders on ARM machines, we were pleasantly surprised to find that our ARM encoders could deliver performance on par with our x86 encoders. We are excited to share these observations. While there is always room for further optimization of our Aurora encoders and AI enhancers on x86, we want to share our findings to correct some misconceptions about encoding performance on ARM. We greatly appreciate our customers' continuous support and trust. Together, we are creating something truly innovative.

7