In this paper we discuss monitoring of the convergence of the Generalized Hebbian Learning algorithm which performs Principal Component Analysis, or the Karhunen-Loeve transform. We start with presentation of the details of the internal structure of the algorithm. Next we consider a number of inter-related convergence conditions. Finally, we present example of image coding algorithm in which those conditions are verified.
|Number of pages||5|
|Publication status||Published - 1 Jan 1998|
|Event||Proceedings of the 1998 IEEE International Joint Conference on Neural Networks. Part 1 (of 3) - Anchorage, AK, USA|
Duration: 4 May 1998 → 9 May 1998
|Conference||Proceedings of the 1998 IEEE International Joint Conference on Neural Networks. Part 1 (of 3)|
|City||Anchorage, AK, USA|
|Period||4/05/98 → 9/05/98|