The LSSVMs can handle thousand and even million sample datasets. But a special method must be implemented.
Fixed-sized LSSVM
The basic idea of the Fixed-sized algorithm is to employ low-rank approximations and an active selection of the support vectors by means of information theory. The full algorithm can be read on the book.
Note that this is also a form of pruning, so it can be used as both for pruning and for large-scale training and prediction.
The LSSVMs can handle thousand and even million sample datasets. But a special method must be implemented.
Fixed-sized LSSVM
The basic idea of the Fixed-sized algorithm is to employ low-rank approximations and an active selection of the support vectors by means of information theory. The full algorithm can be read on the book. Note that this is also a form of pruning, so it can be used as both for pruning and for large-scale training and prediction.