Method of boundary pairs
The Border Pairs Method ( BPM ) is an algorithm for machine learning of the multi-layer perceptron . It is the structure of the bipropagation method , which in turn is a structure of the backpropagation method . It enables us to arrange the various elements in classes. The advantage of BPM over backpropagation is that it is constructive. This means that the BPM finds the structure of the multi-layer perceptron during learning , which is close to the optimal structure and the learning does not get lost in the local minimum . The function of the method has been tested using synthetic (XOR) and real learning patterns (irises, decimal digits). The method also achieves good results with noise data. In addition to normal learning, "incremental" and "online" BPM learning are also possible.
See also
Individual evidence
- ↑ Adaptive and intelligent systems: second international conference; proceedings / ICAIS 2011, Klagenfurt, Austria, September 6 - 8, 2011. Springer Science + Business Media , 2011, ISBN 978-3-642-23856-7 .
- ↑ Border Pairs Method-constructive MLP learning classification algorithm. In: Neurocomputing. Volume 126, February 27, 2014, pp. 180-187.