Intel Labs at Intel® Vision

Highlights:

  • Intel Vision Conference May 10-11, 2022 Gaylord Texan Resort & Convention Center Dallas, TX

  • 6 Vision of the Future sessions on AI and security and 4 proof-point demonstrations supporting session content

  • In-person and on-demand attendance options

author-image

Intel® is pleased to present the Intel® Vision 2022 conference, May 10-11, an event designed to empower business leaders and technology administrators with solution-driven insights for the future.  As a conference participant, Intel Labs is equally excited to bring you the Vision for the Future track, where top researchers from Intel Labs, entrepreneurs from Intel Ignite, leaders from Intel Foundry Services, and academic partners will discuss leading-edge technologies for business leaders and technology administrators. 

Find out how Intel Labs is taking steps today to help businesses and organizations address impending challenges. Sessions will cover the future of AI and security in computing and provide context for additional demonstrations that are included as part of the track. Whether you are interested in the future of AI, security, cloud, hardware, or software, the Vision for the Future track has you covered with sessions that explore the following:
 

  • Security for financial services, including important developments in homomorphic encryption
  • The benefits of human-AI collaboration in manufacturing
  • How computer vision continues to transform retailer and customer interactions
  • How federated learning is paving the way for deep learning collaborations in healthcare
  • Applications of cognitive AI technologies in retail and recommendations systems
  • And much more

Join us and take a proactive step in learning about leading-edge technology and next-generation solutions. To register for the in-person or on-demand Visions event, or to receive more information about other Vision tracks, go to Intel® Vision.  

Following is a complete list of Intel Labs’ presentations and demonstrations at Intel Vision: 

The Future of Security and Privacy in AI/ML 

Jason Martin, Principal Engineer in Intel’s Security Solutions Lab
Session VISI003 on Tuesday, May 10 at 10:45 AM     

A thoughtful, end-to-end approach to security and privacy is a non-negotiable aspect of AI and machine learning. But what does that look like? And what is Intel doing to provide a more trustworthy future, while paving the way for pioneering research in healthcare, science, and business?
Maintaining secure data is not just a matter of software encryption. Hardware systems also play an important role. In this session, you will learn what Intel Labs is doing to protect the future of data, even as it is being shared for the common good.
In this session we explore help the potential vulnerabilities of machine learning and what Intel is doing to shore up security and privacy against future data breaches. You’ll also get an exciting look at Intel Lab’s “adversarial machine learning” initiative and how it is being used to create more robust and secure algorithms.

Neuromorphic Computing: Achieving 1000x Gains in AI at the Extreme Edge

Mike Davies, Director of Intel’s Neuromorphic Computing Lab
Session VISI004 on Tuesday, May 10 at 10:45 AM     

AI systems at the extreme edge such as autonomous robots, drones, and vehicles, require data center-level performance but are constrained by size, data, power, and latency. By applying insights from the human brain, neuromorphic computing is re-inventing the foundational architecture of computers to improve AI performance and power intelligent devices with greater efficiency.
As we advance into this new frontier, we find that the opportunities for neuromorphic technology extend beyond autonomous systems. Already we can demonstrate a broad range of workloads spanning AI, signal processing, optimization, and control workloads where neuromorphic computing can deliver gains of 1000x or more.
In this session, we’ll share the current trajectory of neuromorphic computing, made possible by the recent release of Loihi 2, Intel’s second-gen neuromorphic research processor. This innovation, along with Lava, an open-source software framework, gives greater accessibility to mainstream developers, and faster progress to commercial impact.

Human-AI Collaboration in Manufacturing

Lama Nachman, Director of the Intelligent Systems Research Lab at Intel 
Session VISI005 on Tuesday, May 10 at 1:45 PM

AI deployments, especially deep learning, have spread rapidly through many verticals, including healthcare, manufacturing, retail, entertainment, and others. Amid this growth are rising concerns over job loss, privacy, safety, robustness, and sustainability. These concerns have skewed the narrative on AI to one of liability, rather than opportunity. It’s time to readjust the lens and recognize the many opportunities for human/AI collaboration. 
Humans and AI capabilities are more complementary than overlapping. AI excels at tasks that require broad exploration, analysis, and pattern recognition of massive data. Humans excel at learning from very limited data, transferring their knowledge easily to new domains, and making real-time decisions in complex and ambiguous settings. 
In this session, you’ll learn about the benefits of human/AI collaboration in manufacturing, education, and assistive computing. 

Post-Quantum Cryptography: Defending Against Future Adversaries

Manoj Sastry, Principal Engineer in the Security & Privacy Research Lab at Intel Labs
Session VISI002 on Wednesday, May 11 at 10:15 AM     

Quantum computing leverages the laws of quantum mechanics to solve problems that classic computing cannot. It has the potential to solve many currently intractable problems, giving way to life-changing breakthroughs in chemistry, medical science, and materials science. However, quantum computing also has the potential to disrupt our digital economy by breaking classical cryptographic algorithms that secure our data and communication. 
Post-quantum cryptography is the answer to maintaining security while enabling the vast potential of quantum computing. Based on mathematical problems considered too difficult for even quantum computers to solve, post-quantum cryptography is urgently needed to replace current public-key cryptosystems.
In this session, we will explore the threats posed by quantum computers, post-quantum crypto standards, and what Intel is doing to protect against adversaries in an emerging quantum world. 

Homomorphic Computing: Achieving the Pinnacle of Data Privacy

Rosario Cammarota, Principal Engineer at Intel Labs 
Session VISI001 on Wednesday, May 11 at 1:45 PM

Organizational leaders know that there is tremendous opportunity in the data they own. The challenge is responsibly leveraging that data without exposing it. In this session, you’ll learn how Intel is taking data privacy to the next level with homomorphic computing.
We all know data can be encrypted for privacy but deriving value from it typically involves computations that require decryption while the data is in use. Homomorphic computing allows for computation on encrypted data, thereby eliminating vulnerability. The data and resulting computations remain encrypted until the data owner chooses to decrypt it.
Intel, in collaboration with Microsoft, has developed technology that provides end-to-end data encryption. In this session, you’ll learn how Intel is paving the way for homomorphic computing with key deployments including the Defense Advanced Research Projects Agency (DARPA) so that organizations can begin applying it sooner rather than later.

Cognitive AI: Architecting the Future of Machine Intelligence

Ted Willke, Senior Principal Engineer, Intel Labs, and Director of Intel’s Brain-Inspired Computing
Session VISI006 on Wednesday, May 11 at 1:45 PM

Human-centric, cognitive AI is the future of machine learning. By 2025, machines are expected to make great advances in understanding language, integrating commonsense knowledge, reasoning, and autonomously adapting to new circumstances. 
A key tenet of this evolution is multimodal cognition, the ability for machines to acquire knowledge from a variety of inputs, understand the world, and apply reasoning, thus mimicking how humans learn from their environment.  Multimodal cognition will bring machines one step closer to human-level performance in a variety of real-world applications that demand deliberation.
This session will reshape your vision of AI as one of a co-collaborator with humans as we navigate the many challenges of life and work. .

Demonstrations

AI for Building a Trustful Media Ecosystem
The manipulation of audio-visual media is increasingly rampant, causing social erosion of trust in digital media. In the battle against deepfakes, we demonstrate the world's first real-time deepfake detection platform. With our renowned FakeCatcher algorithm at the core, the AI-accelerated platform uses Intel frameworks and technologies that deliver multiple real-time detection streams.  We conclude with our responsible deepfakes and provenance initiatives to drive a trustful future.

Human-Robot Collaboration in Task Co-execution
Compelling and novel robotics solutions can be created, tested, deployed easily and economically while taking advantage of Intel’s holistic hardware and software. To support this key-message, high-performance AI algorithms capable of robustly addressing real-world problems are required. We tackle these challenges based on Intel’s unique sensing and computing capabilities. Results will be experienced first-hand in an engaging manner through live robot task executions with performance and real-time visualizations.

Mining Knowledge from Video at Scale on Xeon
A new wave of AI is emerging which will lead to higher machine cognition. This will be characterized by new architectures that integrate Neural Networks, Structured Knowledge, Symbolic Reasoning and Broad Information Extraction, leading to robust inherently multimodal machine systems capable of cognition-level abilities in terms of reasoning, conceptualization and abstraction. Such fundamental new capabilities, reflected in areas like vision + language understanding and reasoning, will drive new system architectures optimized for Cognitive AI.

Using Federated Learning to find Brain Tumors
We will introduce Intel’s Federated Learning (FL) framework (OpenFL) running on Intel SGX for training AI models across multiple private data silos without requiring sharing of data. We will present the first large scale real-world FL study, involving data from n = 71 healthcare institutions across 6 continents with the University of Pennsylvania