Abstract:

Constructing reasonable directed acyclic graphs is a key problem that is to be solved urgently. An improved directed acyclic graphs support vector machine is presented, which obtains the minimal hyperspheres according to the hypersphere support vector machine, computes the minimal distances among classes according to the minimal hyperspheres, gets the minimal distance matrix according to the minimal distances, and constructs the directed acyclic graphs according to the minimal distance matrix. The experimental results show that the algorithm has a higher classification precision, compared with the old directed acyclic graphs support vector machines.

Key words: directed acyclic graphs support vector machine(DAGSVM);hypersphere support vector machine;minimal distance matrix