Presented by Qualcomm

AI is revolutionizing industries, products, and core capabilities with dramatically enhanced experiences. Learn about the breakthroughs in AI research that are pushing the boundaries of what’s possible and shaping the future of AI when you join this VB Live event!

Register for free here.

AI use cases are expanding beyond big data and out of the cloud. Thanks to the increasing sophistication and processing capacity of modern devices, artificial intelligence is moving to the edge, powering up mobile, machines, vehicles, and connected things across the spectrum and in every industry.

That means we’re also seeing a level up in the capabilities of AI, from computer vision to natural language processing, from the medical field — where an AI system can detect melanoma better than human doctors — to robotic vacuums smart enough not to fall down the stairs, instant translation on our devices, and more.

Where AI works, it works far better than any human capability could manage; however, its capabilities remain surprisingly restricted as potential applications become more complex and start requiring things like reasoning. For instance, image analysis and speech recognition are incredibly sophisticated — an algorithm can identify objects in a scene, but couldn’t put all those elements together to determine why something is occurring, or what might happen next.

For applications like a perfectly safe self-driving car, that could be fatal — the automobile’s AI needs to not just be able to process signals from other drivers and about pedestrians and various obstacles; it needs to understand and anticipate the intentions and possible outcomes.

And that’s just one of the many arenas of research data scientists are applying themselves to. In order to realize the true potential of artificial intelligence and machine learning, to pursue even more sophisticated applications and more of that fascinatingly near-miraculous functionality, scientists are getting busy on various fronts.

Advanced neural network approaches: The goal of most AI and machine learning is related to teaching machines to think, reason, and make decisions like human beings do. Neural network approaches semi-supervised machine learning, which offers an algorithm a small batch of labeled data to extrapolate from, mimics human learning.

Generative adversarial networks (GANs) also have huge potential, and have already demonstrated an ability to mimic any distribution of data — or in other words, create images, music, speech, prose, and more that’s incredibly similar to our own. Distributed and cooperative learning systems, including federated learning systems, enable true learning and are designed for real-world, edge-based AI, because they don’t require the major computing power of the cloud to run.

Cutting edge hardware: Nothing is more important in AI than the hardware it is run on. But this generation of deep learning algorithms is getting close to hitting the ceiling, in terms of energy consumption, because they’re energy hungry and deeply inefficient. Researchers are working on breakthroughs in low-power hardware that can handle these machine learning workloads.

Network optimization: The opportunities for always-on, intelligent devices that do all or most of their thinking on the device are enormous, and to make that happen, networks need to level up, meaning research is needed into compression, inter-layer optimizations, optimizations for sparsity, and other techniques to take better advantage of memory and space/time complexity, to provide highly responsive, highly secure, and intuitive user experiences.

Chip and algorithm co-dependency: Here’s where machine learning meets hardware, and creates something brand new. Researchers are looking for ways to train an algorithm to perform well on a chip with limited memory and precision — and conversely, a chip that can execute deep learning algorithms efficiently and at low power.

There’s a tremendous amount of space and potential for innovation in the artificial intelligence and machine learning field, and researchers are taking major leaps forward, racing toward a completely connected, intelligent future of sophisticated hardware and always-on AI. The number and variety of potential use cases across industries is growing as fast as the ideas for applications — and our ability to execute those ideas — is growing.

To learn more about what the minds of the brightest AI scientists are accomplishing, the problems they’re solving, and the solutions that are changing the world, don’t miss this VB Live event!

Don’t miss out!

Registration is free here.

In this webinar, we’ll discuss:

  • Several research topics across the entire spectrum of AI, such as generalized CNNs and deep generative models
  • AI model optimization research for power efficiency, including compression, quantization, and compilation
  • Advances in AI research to make AI ubiquitous


  • Jilei Hou, Senior Director, Engineering, Qualcomm Technologies, Inc.
  • Jack Gold, Founder & Principal Analyst, J. Gold Associates