Sunday, October 19, 2014



Partial Project Summary History
Project descriptions
Technologies and skills
Signal/Image Processing and Pattern Recognition

Tank image recognition from satellites
3-D stereoscopic extraction algorithms
Submarine signature recognition from acoustic underwater phased arrays
Expert Bayesian system, parameter tuning, phased array signal processing correlogram
Stock market statistical arbitrage
Creating models based on financial fluid flow dynamics and correlations of stock prices. Forecasting (Tuck).
Credit card fraud pattern recognition.
Spending pattern irregularities alerts
Sound triangulation/sniper system
Asymmetric phased arrays and reflection calibration algorithms
Biometric recognition systems

Face, finger print, retinal, voice and infra-red vascular recognition
Feature extraction, registration algorithms, noise removal, large dataset management. Eigenvectors, wavelets, 3-D convolutions, cluster analysis.
Eye gaze tracker
Real time image processing. Head rotation modeling.
Truth detection systems
Automating polygraph parameter sensors and voice stress. Emotional context modeling through vocabulary templates.
Gesture recognition
Hand gestures conversion to computer commands
Brain Print Recognition
Utilizing many samples, human brains have a unique MEG/EEG brain “hum” and cluster analysis can be done to recognize mental illnesses and uniqueness of individuals. (Lockheed Martin is doing research on recognition of precognitive signals of objects. Berkley also just released a paper showing how to discriminate between 1024 visual objects.)
Supercomputing Applications

Various projects including FFT optimization
Vector Pipeline, BBN Butterfly, Hypercube, Viking Boards
Artificial life simulations, gas lattices
Specialized Hardware, genetic and evolutionary algorithms.
Grid computing
BEA and Oracle distributed grid computing
Computer generated holography
3-D spatial FFT on BBN’s connection machine. C-paris parallel programming language. Fortran 77.
Speech Recognition/Identification

Voice recognition/identification
BBN Hark’s system.Markov modeling, n-grams, context probability models, speaker independent models
Models of discourse and planning
Studied under world renown Barbara Grosz, graduate school adviser Harvard
Natural Language parsing and AI knowledge networks
CIA project. Focused on pronoun reference algorithms.
Biomedical Projects

Soldier 2000 project
Connecting vital sensors to a wireless bodyLan . DARPA.
Robotic Surgery in Medicine for the Army
Haptic feedback, latency time compensation,  tremor filtering, object touch recognition
Virtual Reality in Medicine
Overlays of fMRI scans onto patient for real time surgical applications
Virtual Reality 3-D Reconstruction Teleconferencing
Carnegie Mellon University. 3-D reconstruction of objects in the teleconference room.
EEG controlled robots
Relatively simplistic mental state recognition to control direction of robot. Discriminant function analysis.
Neural network controller for “INFANT” robot
Designing robot controllers to learn to grasp objects that mimic human babies. Mentor: Dr. Kuperstein (MIT)
Retraining the brain to interpret tactile feedback as visual information
Brail interfacing also used by Navy for Sonar feedback on tongue tactile vision systems.
Research / simulations in cybernetic hive mind work groups
(BBN) Increasing the hypothetical efficiency of integrated rapid communication technologies.
Neural network control system for leg muscle stimulation to make paralyzed people walk again
Engineering of extoskeleton. Control systems for non-linear biological models due to muscle fatigue.
Digital circuit / systems design

Cascadable digital neural network chip
VLSI design, production, testing
Wearable ubiquitous computing systems
Enhanced reality, enhanced cognition. Heads up display borrowed from NASA for entrepreneurial venture.