Le projet THINK

Capitalize on distributed architectures based on Kubernetes* with Intel® optimized AI software and hardware.

Highlights from Ripan Roy blog proposing a solution to Face Mask Detection

Learn about Sign Language Recognition technology that recognizes actions from sign languages

Take advantage of mixed precision to make DL models run faster during training and inference

The International Conference on Learning Representations (ICLR) 2023 will run from May 1st through 5th in Kigali, Rwanda. Intel Labs’ innovations in model linearization include a three-stage training method that trains a DNN model with significantly fewer rectified linear units (ReLUs) driven by a novel measure of non-linearity layers’ ReLU sensitivity.

Using SYCL Joint Matrix Extension on Intel® AMX and Intel® XMX to create custom AI optimizations

Intel Labs excitedly presented its recent research, ‘Multi-Modal Conversational Artificial Intelligence in Early Childhood Education: Kid Space at Home’ at British Educational Training and Technology United Kingdom (Bett UK 2023) in March.

In this blog, we introduce the challenges of traditional NAS solutions and propose a multi-model, hardware-aware, train-free NAS to resolve those challenges. It includes a unified transformer-based search space, a hardware-aware search strategy, and a train-free score method. We also present how this solution improves the multi-domain models’ performance on commodity CPU clusters.

In this blog, we evaluate the progression of methods to guide and converse with machines based on recent technology innovations, such as OpenAI’s ChatGPT and GPT-4, and explore the next needed steps in conversational AIs toward mastering natural conversation like a human confidant. Machines have made a categorical leap from prompt engineering to “human speak,” however other aspects of intelligence are still awaiting discovery.

Intel Labs, in a joint effort with Brown University and MIT through a DSAIL research collaboration, has been crowned as the new record holder for the JouleSort “Indy” Category, completing the task using just 63 kJ.