Scott Bair is a key voice at Intel Labs, sharing insights into innovative research for inventing tomorrow’s technology.
Highlights:
- Intel’s Innovation Selects features a collection of specially curated technical talks and demos, including product deep dives, tailored for developers and tech enthusiasts.
- The collection highlights groundbreaking work from Intel Labs in artificial intelligence, including new open-source tools and open benchmarks and progress toward human-centric cognitive AI.
Intel’s Innovation Selects features a collection of specially curated technical talks, demos, and product deep dives, each tailored for developers and tech enthusiasts. Content covers a wide range of topics designed to help technical experts and enthusiasts stay ahead and informed of the ever-evolving tech landscape.
As part of the collection, Intel Labs is proud to present groundbreaking work in artificial intelligence. Learn from our researchers about how Intel Labs is pushing the state of the art in AI by establishing open-source tools and new open benchmarks for industry and academia. Listen as they teach about how human-centric cognitive AI is the future of machine learning.
Intel Labs’ work was also highlighted in a demonstration that compares the Intel Scalable Vector Search library to the widely used open-source vector search library HNSWlib. Results show a latency improvement of over eight times compared to HNSWlib when using 45 million vector embeddings with 1536 dimensions.
Beyond LLMs: Foundation AI Models of Tomorrow
Presented by Dr. Somdeb Majumdar, the Director of AI Lab – a research organization within Intel Labs. He received his Ph.D. from UCLA and spent several years developing ultra-low-power communication systems, wearable medical devices and deep learning systems. He has been published in top-tier journals and conferences and holds 27 U.S. patents. At Intel Labs, he leads a multi-disciplinary team developing foundational AI algorithms, scalable open-source software tools and disruptive applications in Computer Vision, Chip Design, Graph Learning, Scientific Computing and other emerging areas.
This presentation details Intel Labs' groundbreaking work in AI and how our researchers are leading the AI open ecosystem in solving problems in several domains. Large language models (LLMs) have transformed the public perception of AI. Intel Labs is already thinking about the next big frontier. Somdeb explores the scaling behavior of LLMs and how to incorporate them in consumer-grade hardware. He also tackles open questions like how to better ground complex reasoning systems in knowledge and build disruptive solutions for areas like computational chemistry, chip design, and video understanding. Watch the full video to learn more about these research vectors and how we are pushing the state of the art in AI to establish open-source tools and new open benchmarks for industry and academia.
Cognitive AI: Multimodal RAG on Intel® Gaudi® Accelerators
Presented by Vasudev Lal, an AI Research Scientist at Intel Labs, who leads the Multimodal Cognitive AI team. His team develops AI systems that can synthesize concept-level understanding from multiple modalities: vision, language, video, and audio. His current research interests include equipping deep learning with mechanisms to inject external knowledge; self-supervised training at scale for continuous and high dimensional modalities like images, video and audio; and mechanisms to combine deep learning with symbolic compute.
Cognitive AI is the future of machine learning, aiming to bridge the gap between human and machine intelligence. Multimodal cognition, a key aspect of this evolution, enables machines to learn from various inputs and to reason like humans. Intel® Gaudi® 3 AI accelerators are designed to power this future, efficiently handling the demanding workloads of multimodal AI.
Watch the presentation below, where Vasudev details how human-centric, cognitive AI is the future of machine learning. By 2025, machines are expected to advance in understanding language, integrating commonsense knowledge, reasoning, and autonomously adapting to new circumstances. A key tenet of this evolution is multimodal cognition — machines will gain knowledge from a variety of inputs to understand and apply reasoning like humans. Multimodal cognition will bring machines closer to human-level performance in a variety of real-world applications. Learn how Intel® Gaudi® 3 AI accelerators are built to handle demanding training and inference for multimodal AI. Watch a demo through Intel® Tiber™ Developer Cloud on how to save time and power.
Intel® Scalable Vector Search on the Intel® Xeon® 6 Processor
Demo
In the deep learning era, high-dimensional vectors have become the quintessential data representation for all forms of unstructured data. Searching for semantically similar vectors is a critical task for many modern applications. They include semantic search during web search, recommender systems, and emerging applications, notably retrieval augmented generation (RAG) for factual grounding, long-term memory, and hallucination mitigation in AI systems.
Intel Labs has developed an optimized Scalable Vector Search (SVS) performance library for vector similarity search. Using the latest generation of the Intel® Xeon® 6972P Processor (96C), the following video compares the Intel SVS library to the widely used open-source vector search library HNSWlib, showing a performance benefit of over 8x in latency over HNSWlib when using 45 million vector embeddings with 1536 dimensions.
On HNSWlib, vector search starts to backlog when there are more than 25 thousand concurrent users, slowing down the chatbot response or making it unresponsive. Intel® Scalable® Vector Search (SVS) can run over 200,000 concurrent users with the same configuration.
The demo also compares the similarity search performance using Intel SVS on both AMD* EPYC* 9654 Processor (96c) and the Intel® Xeon® 6972P Processor (96c). Platforms can take advantage of this library, and Intel hardware will take advantage of the Intel-optimized features such as vector compression. This example shows the superiority of Intel hardware plus Intel-optimized software on this workload. The Intel® Xeon® 6 platform similarity search performance is 6.8x faster than AMD Genoa at the same core count. One would need various Genoa servers to deliver the same vector search capabilities that can be completed by a single Xeon 6 system.
Performance claims used can be found here.
Scott Bair is a Senior Technical Creative Director for Intel Labs, chartered with growing awareness for Intel’s leading-edge research activities, like AI, Neuromorphic Computing and Quantum Computing. Scott is responsible for driving marketing strategy, messaging, and asset creation for Intel Labs and its joint-research activities. In addition to his work at Intel, he has a passion for audio technology and is an active father of 5 children.
Scott has over 23 years of experience in the computing industry bringing new products and technology to market. During his 15 years at Intel, he has worked in a variety of roles from R&D, architecture, strategic planning, product marketing, and technology evangelism. Scott has an undergraduate degree in Electrical and Computer Engineering and a Masters of Business Administration from Brigham Young University.