Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.
I have been studying Data Mining techniques and algorithms for 3 years already, therefore I possess enormous experience in Data Mining Applications to DSP. I'll just enumerate them here in order not to shrink place for other participants. If anybody is really interested in, just contact me. I'll be looking forward to.
1) Segmentation of signals: Spliting the signal into fragments of generally different sizes (lengths), which possess homogeneous properties. Segmentation helps finding out pieces of signal with different properties (spectral, ststistical, informative, etc.)
2) Clustering - Uniting the discovered segments into a fixed number of classes (taxones, clusterers) which is defined deliberately or automatically. The unification is done according to the special algorithms (k-means, k-median, EM, CobWeb, AVP ,etc)
3) Mining Assosiacion Rules and Sequential Analysis - This direction consists od finding logical regularities in signal's internal structure and helps connecting different events in signals. Also it includes finding time patterns (sequential analysis) which also characterize the signal's nature
4) Classification - Referring new signals or separate fragments to one of the known classes according to the constructed model (trees, rules, graphs, etc.)
5) Multi-dimensional visualization - very suitable and convenient for plotting n-dimensional dataset in order to observe their proximity or remoteness which is useful for explaining their classes labels.
Recently, Statistical analysis has been involved in the group of DM problems. However, it's usually effective in preliminary processing, when absolutely nothing is aware about the dataset