Massive Data Sets Proceedings of a Workshop (1996) / Chapter Skim
Currently Skimming:
Massive Data Sets in Navy Problems
Massive Data Sets in Navy Problems
Pages 157-168
The Chapter Skim interface presents what we've algorithmically identified as the most significant single chunk of text within every page in the chapter.
Select key terms on the right to highlight them within pages of the chapter.
From page 157... ...
S Navy that, by their very nature, fall within the realm of massive data sets.
|
From page 158... ...
Helicopter ~ 17 -IU ~ 0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 Figure 2 - Grayscale images of a tank and mammogram. fielding an instrument which will be capable of classifying acoustic signatures under battlefield conditions in as near to a real-time manner as possible.
|
From page 159... ...
In the following sections we discuss at length the specific difficulties that we have encountered analyzing the data sets associated with these two applications. There are associated problems with both the exploratory data analysis of potential features, the development of density estimates of these features, and the ultimate testing and fielding of classification systems based on these features.
|
From page 160... ...
3.5 2.5 1.5 0.5 3 1 1 O . X 105 0 10 20 30 40 50 60 70 Figure 3 - Power spectral density of one of the helicopter signals.
|
From page 161... ...
and O(n2) depending on the nature of the algorithm, extremely long execution times can be expected for the analysis of these large data sets.
|
From page 162... ...
In this case the huge data sets that are associated with each class can lead to extremely slow convergence of recursive density estimation procedures. This phenomena has been observed by us in the case of likelihood maximization of over determined mixture models under the Expectation Maximization (EM)
|
From page 163... ...
These packages are aimed at providing a seamless interface between statistical analysis software and database management systems. That being said' there are yet fundamental research issues associated with the seamless integration of statistical sof~vvare and database management systems.
|
From page 164... ...
~., There remain many technical issues associated with probability density estimation for massive data sets in moderately high-dimensional spaces. Efficient procedures for estimating these densities and performing cluster analysis on them are needed.
|
From page 165... ...
In addition research into the use of pseudo-metrics such as Kullback Leibler for comparisons of probability density functions as an alternative to the standard likelihood ratio hypothesis testing deserves continued research. Finally procedures that attempt to combine some of the modern density estimation techniques with the Markov random field approach have merit.
|
From page 166... ...
(1995) "A recursive deterministic method for robust estimation of multivariate location and shape," accepted pending revision to the Journal of Compulafional and Graphical Stafisfics Poston, W.L.
|
From page 167... ...
(1990) "Hyperdimension~ dam analysis using parallel coo~ina~s," ^~' e 4~_~n I 837 6~-~.
|
Key Terms
This material may be derived from roughly machine-read images, and so is provided only to facilitate research.
More
information on Chapter Skim is available.