Abstract

Analyzing large image and text databases poses particular computational problems. Computational problems can sometimes be solved by using traditional analysis techniques, and by throwing more and more memory cycles at them. A more aesthetic way to tackle such scalability problems is to find new data structures and new algorithms which will more thoroughly deal with these issues. One of the most looming issues in data analysis is the laborious phase prior to the main analysis: selection of data, coding, etc. We summarize some recent results in data coding. We then look at how the incorporation of the wavelet transform into data analysis can helpfully mitigate some problems related to preliminary data processing. We look at how these same principles (but with a different wavelet transform) can be used in time series prediction. 1 Preliminary Data Processing, Coding and Analysis: Overview The major part of data analysis does not go into the analysis itself, but rather into the demarcatio...

Citations