NVIDIA Latest Push in Generative AI the Metaverse
Sundays, The Sequence Scope brings a summary of the most important research papers, technology releases and VC funding deals in the artificial intelligence space.
🗓 Next week in TheSequence Edge:
Edge #259: Our series about ML interpretability continues by exploring the SHapley Additive exPlanations(SHAP) method, reviews MIT’s research in a methodology for classifying interpretable features in ML models and evaluates the iModels framework from Berkeley University.
Edge #260: Deep dives into Meta AI’s recent Data2vec 2.0 self-supervised model for speech, vision and text.
📝 Editorial
The Consumer Electronic Show(CES) captures the tech headlines the first week of every year. For the last few years, AI has been front at center of CES but most of the highlighted focuses on, well, robotics and electronics. NVIDIA is one of the few tech incumbents that has achieved a leader positions in AI research, hardware-electronics as well as software. NVIDIA took full advantage of CES 2023 to unveil some of their latest AI efforts several of which, not surprisingly, are centered around generative AI.
The most exciting NVIDIA announcements came from the addition of generative AI 3D technologies to the Omniverse platform. NVIDIA Omniverse has become one of the most complete platforms for building metaverse applications and generative AI is now one of their key cornerstones. Omniverse now includes connectors to lead generative AI platforms such as Move.ai for body movements, Lumirithmic for facial 3D meshes or Elevate3D for photorealistic 3D visualizations. NVIDIA also announced Get3D, a new generative AI model that can create 3D shapes with topology. That and many other interesting generative AI capabilities of the Omniverse platform are now available at the Omniverse AI ToyBox extension platform.
NVIDIA has one of the broadest spectrums in the world for deploying generative AI applications at scale. Omniverse is just an easy way to start.
🔎 ML Research
Video Annotations
Amazon Science published a paper proposing a human-in-the-loop semantic segmentation method for labeling video data —>Read more.
Graph Machine Learning
Hugging Face published a detailed blog post with an introduction and research summary about the most important concepts in graph machine learning —>Read more.
Image and Text Understanding
Researchers from University of Oxford and LMU Munich University published a paper proposing a framework machine understanding of images and text using generative models —>Read more.
🤖 Cool AI Tech Releases
NVIDIA Omniverse and Generative AI
NVIDIA announced the addition of generative AI capabilities that include the creation and animation of 3D objects and modifying characters —>Read more.
TextBox 2.0
A new version of TextBox, a text generation library based on PyTorch has been released —>Read more.
🛠 Real World ML
Accelerating PyTorch Training
Engineers from Hugging Face discuss how to accelerate PyTorch training jogs using a cluster of Intel Rapid servers on AWS —>Read more.
💸 Money in AI
Data Cloud powerhouse Snowflake announced the acquisition of time series forecasting platform Myst.
Taiwanese startup Profet AI raised $5.6 million to building prediction models and intelligent applications for industrial manufacturers.
Employee productivity startup Prodoscore announced a strategic growth investment led by PE firm PSG.
Real time data streaming, and real time ML infrastructure darling, Confluent is acquiring Immerok who provides managed services for Apache Flink.
The link regarding image and text understanding from the University of Oxford and LMU is incorrect. It simply points to TextBox 2.0, which is a separate topic covered in the article. Kindly post the correct link here and either update the article or make a correction in a future article. I am quite interested in this topic. Thank you kindly!