Well, the amount of techniques is really enormous. Everything depends on the concrete goal you want to achieve with your classification. I'll mention several groups of algorithms and afterwards you'll choose the necessary ones according to your task. In general, all these methods refer to the so-called Data Mining, which tries to discover new, non-trivial, practically useful and adjusted for human's interpretation knoledge just from initial data sets.
1) Segmentation algorithms. Segmentation divides the signal (the data set) in to a number of fragments with different length, each of which possesses the uniform characteristics. Usually segmentation is applied during the step of preliminary exploration of signals in order to find out the boundaries of sements. It's considered that different segments correspond to various types of processes (noise, trend, transmission, etc.)
Segmentation algorithms: 1) Autoregressive models
2) Principal component segmentation
3) Statistical segmentation (AIC,BIC, MDL criterions, etc).
4) Cumulative sum segmentation
5) Segmentation based on smoothing
and others.
2) Clustering. Cluster-analysis is used in order to find out the segments , which are turned out to refer to the same classes, hence, they posses the same or very close properties (statistical, spectral, etc.). Clustering is used after segmentation. The majority of algorithms is based either on computing the centres of clusterers and combining the vectors, which represent the same class, or on automatical clusterisation, when the best and most suitable number of clusterers is established adaptively. The most popular algorithms are:
1) k-means, k-median
2) EM
3) FartherstFirst
4) CobWeb
5) Hierarchical clustering
6) Statistical clustering
7) Graph models
3) Classification
Imagine you've performed clustering. Then you get a new fragment (or even a new sample) and you need to find out, which class it belongs to. This problem is solved by powerful algorithms of classification. The results may be expressed in rules and trees. The algorithms:
1) J48
2) Naive Bayes
4) Sequential analysis
This direction in Data Mining tries to sort all events according to their time arisal. This denotes, that it searches the so-called time-patterns, which are existed in 13 forms (according to Allen's logics). Sequential analysis helps answer the question which patterns are earlier or later than the others
_______________________________________________
I really advise you to get acquanted or to read any book about Data Mining
With respect,
Dmitrij