Visual Search using Vector Quantization

On Wednesday, 14 January 2015, Ole Eichhorn will discuss Visual Search using Vector Quantization.

Abstract

The techniques of vector quantization have long been used for data compression, but they can also be used as an innovative way to accomplish pattern recognition. When a dataset is “kernelized” it is transformed into a smaller denser space that enables fast lossy searching. This can be particularly useful as a way to implement image matching for visual search. This talk will describe various ways of “kernelizing” images and methods for storing and retrieving the resulting vectors. It will also describe a visual search engine implementation based on these methods, and touch on related techniques including kernel synonymization and iterative turning. And yes, there will be a demo.

About the Speaker

Ole Eichhorn has more than 25 years of experience as an entrepreneur, software engineer, architect, and manager.  In 2013 he founded eyesFinder, a company developing leading-edge Visual Search solutions.  In 2011 he co-founded DoubleBeam, which develops mobile payments solutions.  Ole was the founding CTO of Aperio in 2001, the leader in digital pathology solutions, and actively participated in developing applications of this technology for patient care.  Previous to that Ole served as SVP/Engineering at PayPal, GM of Online Billpay at Intuit, VP/Engineering for Digital Insight, and Director of Technology for XP Systems.

Meeting Info

Meetings are free and open to the public.

Pizza and networking:  6:30 PM
Presentation, Q&A:  7:00 PM

Location:
Richter Auditorium, Ahmanson Science Center
California Lutheran University
Thousand Oaks, CA  91360

Parking:  Do not park in the faculty/staff lots.  Free parking is available in the visitor lot at the corner of Olsen Road and Mountclef Boulevard.  Alternatively, CLU Public Safety has provided us with parking passes for on street parking.

Additional maps:  http://www.ieee-bv-cs.org/meetings/

PDF24    Send article as PDF