A SOM-based probabilistic neural network for classification of ship noises
Jie Chen, Haiying Li, Shiwei Tang, Jincai Sun · 2003
A probabilistic neural network (PNN) is applied to the classification of ship noises for its simplicity in the training process. However, a main limitation of PNNs is that all computations are carried out at runtime, and it may become an overburden if the training set is large. This paper presents a modified PNN algorithm, based on self-organizing maps (SOM), which can reduce the running time through real-time optimization of the training set, and retains the virtue of the training procedure as a simple forward computation at the same time. Experimental results verifying the proposed algorithm are provided.