Basic Research in Computer Vision:Object Recognition, Face Recognition, General Pattern Recognition, Evolution of Evolvability, Evolution of Modular Design, Self Organization and Organic Computing.Targeted Research in Computer Vision and Text Analysis: Commercial (advertisement) Segmentation and Recognition, Multi-Ad Campaign Placement Optimization for TV Media Buys, Topic Segmentation, Sentiment Analysis and Text ClusteringProof of Concept and complete Research and Development write-ups in a Language Independent Journal Format.------------------------------Ace Metrix Project Titles--------------------------------Fingerprinting Audio Recognition System Based on Ideas from Computer Vision and the Brain (Patented)-Real Time Automatic Demographic Smoother For Online Survey Management (Patented )-Commercial Features as a Vector Space-Television Ad Acquisition System-Multi Ad-Campaign Optimization-Text Analytics (document level vector embedding)-Video Recognition System-Video Compression as a means to a robust video representation for database storage and similarity search (key frame selection)-Automatic Video Summary (scene selection for abridged video)-Survey Agreement-Control Analysis and Scores-Logo Detection (in development)-Automatic Advertisement Harvester-----------See Projects Section for more Details----------

SEM Bidding optimization for 20,000+ advertisers. SEM keyword discovery through statistical analysis of millions of user search queries from Google and Yahoo/MSN.

Bayesian modeling and computation of lead conversion rates (with full posterior distribution - not a point calculation) for 20,000+ advertisers at the geo-category level. General on-call problem solver for SEM team.

Design and implement algorithms for text analytics, automation, pattern recognition and optimization.

Implemented a facial detection system in C++ that utilized important aspects of what is known about early stage human visual processing in the primary visual cortex V1. This representation was optimized through evolutionary algorithms and a Bayesian statistical analysis of facial feature vectors. The Bayesian analysis yielded an entropy measure that weighted feature vectors according to relative uniqueness.

Manufactured extremely thin metal films (typically around 20 atoms high) in ultra high vacuum (better than 10^(-11) Torr) and at liquid helium temperatures (4.2 Kelvin). Studied the conduction electron dynamics (magneto-resistance) for insights into magnetic moment formation, Kondo effect, impurity scattering, weak localization, low temperature superconductivity, metal-metal interface effects, spin currents, spin/charge waves, Anomalous Hall Effect.