Deep Learning for IoT : Is there a shallow end of the pool?
Session Abstract: Virtual Only
Deep Learning is thus far a tale of two stories. The more publicized story is one of its step change performance that has astounded even longer-term term practitioners in the field. In trend prediction in IoT, and in face recognition and visual classification, Deep Learning hasn’t beaten the competition so much as crushed it.
The dark side of Deep Learning is the lack of a ‘shallow end of the pool’ for IoT big data practitioners intending to dip their toes into it as an addition to their predictive analytics toolbox. Even a tentative foray into Deep Learning involves choosing between rapidly evolving (and competing) frameworks, and making non-trivial design choices (data sufficiency, data augmentation, network topology, guards against too slow or too fast learning rates to name a few) that are more art than science.
The goal of this session is to provide a practical overview the concepts in deep learning and its industrial applications. We will then cover key elements of TensorFlow, Google’s deep learning framework that provides a breadth of capabilities needed by most practitioners. Finally, we will discuss best practices in algorithm setup and customization that maximize chances of success in leveraging Deep Learning for predictive analytics applications.
This talk is targeted at beginner/intermediate audiences with executive or architectural oversight, in organizations looking to explore Deep Learning technologies for domain problems.
Venu Vasudevan, PhD
Chief Technologist at next.io
Venu is a technology entrepreneur with track record of successful products at the intersection Big Data, AI and embedded cyber-physical systems. His track record in successful predictive analytics based products include machine learning based product that was acquired by Watchwith/Comcast (https://goo.gl/Xf4PNa), a healthcare asset tracking product at Motorola that led to much of Zigbee, and an AI based network management system for Iridium. His work has been featured in Wired magazine and BBC, and he has been a participant and keynote at the Yankee Group, CES, Digital Hollywood venues, as also past speaker at Iot Slam.
He is currently Chief Technologist at next.io, a company developing visual recommenders based on deep learning. He was previously VP of Data Science at Lightpad, a leading consumer IoT startup in intelligent lighting. Venu holds a Ph.D. (Databases & Artificial Intelligence) in Computer Science from The Ohio State University, and is an adjunct faculty at Rice University.
Artificial Intelligence, Machine Learning, Deep Learning, Enterprise, Cloud
VP / Director, Middle Management, Technical
Retail, Manufacturing, Telecom, Industrials
Join our IoT Community at https://www.linkedin.com/groups/4662022/profile