Technomancer wrote:Developed a new extension to a supervised clustering method and applied it to the automatic recognition of active sonar returns. The idea was further taken up by some people in the mine countermeasures group.
I've also developed a novel ICA method based on the Unscented Kalman filter. This worked, but proved to be completely impractical for real problems.
in english please
Technomancer wrote:Umm..I did some sonar stuff for the navy. The I did some hearing aid stuff, it worked but just not for real problems. I also got to present the idea at a conference in Lake Tahoe.
Technomancer wrote:I was with DREA (Defence Research Establishment Atlantic) which is located in Dartmouth, NS.
What was the work you did with Kalman filters?
Also, an announcement! I have now been published (along with many co-authors)! An article entitled "Development of a Realistic Hearing in Noise Test Environment (R-HINT-E)" by Laurel Trainor, Ranil Sonnadara, Karl Wiklund, et al. is to appear in a forthcoming special issue of Signal Processing. My own contribution was the data collection/analysis and the actual MATLAB software for the test environment.
Technomancer wrote:Thanks!
I'm not familiar with the algorithm you mentioned. What was the nature of the problem? There are other tracking/prediction algorithms available. If your problem is non-linear or non-gaussian in nature, you may want to look into Unscented Kalman filtering, or maybe particle filters (or other forms of Bayesian estimation)
Users browsing this forum: No registered users and 116 guests