Research Activities

• EMG Pattern Recognition - Biolectric Control, NASA Press Release, ABC WorldNews Tonight.
• EEG Pattern Recognition - This project aims to improve performance of NASA missions by developing brain-computer interface (BCI) technologies for augmented human-system interaction. BCI technologies will add completely new modes of interaction, which operate in parallel with keyboards, speech, or other manual controls, thereby increasing the bandwidth of human-system interaction. The research will extend recent feasibility demonstrations of electromyographic (EMG) methods for neurocontrol to the domain of electroencephalographic methods or EEG-based neurocontrol. These methods will bypass muscle activity and draw control signals directly from the human brain. BCI technologies will provide powerful and intuitive modes of interaction with 2-D and 3-D data, particularly for visualization and searching in complex data structures, such as geographical maps, satellite images, and terrain databases.
• Intelligent Data Understanding - The Intelligent Data Understanding project is researching and developing machine learning algorithms designed to work with streaming Earth science data.The developing framework has been designed to allow users to graphically program sophisticated algorithms which use data distributed on different machines. The goals of the analysis are to extract the interesting
  components of the datasets to reduce the amount of transmitted data, to automatically diagnose the instrumentation, and to provoke more science questions. The algorithms are being developed as part of a larger framework incorporating data flow programming, interactive graphics, and support for third party software as a plug-in module such as Matlab.
• Biologically Inspired Algorithm - Through out history people have taken inspiration from nature and sythesized practical implementations. The history of artificial neural networks is an example of this process. We know from nature that very sophisticated pattern recognition is performed reliably and quickly with slow and noisy systems. In this project we are looking at the current state of machine learning algorithms and modifying the assumptions which cause a departure from nature. For example many pattern recognition systems assume stationarity along with Gaussian priors. We are investigating techniques which are inherently suited to non-stationary, non-Gaussian systems.

Other

• Biography
• Resume
• Publications
• Contact
  

Links

• NASA Ames
• Neuro Engineering Technical Area
• IFC
• EHS
• SHAS
• IP