Neural ½Òµ{ ±Ð¾Ç¸ê·½
¡i«ö¥kÁä¤U¸ü¡j
¡¯
ÁÈ¿ú¤£®e©ö¡A¼g®ÑªÖ©w§óÃø¡C
¥H¤UÀu¨}°Ñ¦Ò®ÑÄy¡A§Æ±æ¦P¾Ç¦h¦h¿ïÁÊ ¡¯
¡¯
¦n¦n§â´¤¥ú³±¡A×¾i«~¼w¡A¥R¹ê¦Û¤v¡A§µ¾i¤÷¥À¡A»P§g¦@«j ¡¯
²
Matlab «n«ü¥O¡G
>>
help nnet %
Neural Network Toolbox
>>
help nndemos % Neural
Network Demonstrations.
>>
nntool
% ¶i¤J¤@ GUI µe±¡Aª½±µ¶i¦æ¾Ç²ß¡A³o³¡¤À¥i°Ñ¦Ò ¡u°Ñ¦Ò®Ñ¥Ø1¡B3¡v
>>
nnd % ¶i¤J¤@ GUI µe±¡Aª½±µ¶i¦æ¾Ç²ß¡A³o³¡¤À¥i°Ñ¦Ò ¡u°Ñ¦Ò®Ñ¥Ø2¡v
²
³Ì²³æªºNeural
Ntwk. ¡]2_input 0_hidden_layer 1_
neuron 1_output Perceptron¡^ perceptron21.m
½Ð¾\Ū C:\MATLAB7\toolbox\nnet\nndemos\demop1.m µ{¦¡¡C¥ç¥i¦b Matlab ³nÅé >>demo => Neural Network => Perceptrons => Classification
with a 2-input perceptron ¤¤§ä¨ì¡AMatlab
¤]±N¦¹µ{¦¡¼g¦¨¤@Ó GUI µe±ªº§ë¼v¤ù±Ð¾Ç³á¡C
¡i½Ð¦b Matlab ¤¤°õ¦æ
>> nnd2n1 ¡A¨ä¹ê³o¤~¬O³Ì²³æªºÃþ¯«¸gºô¸ô¡]1_input 1_layer 1_ neuron 1_output¡^¡C¡j
¡i¥»¨Ò¬Û·í©ó >> nnd2n2 ¡j
²
¨âÁû¯«¸g¤¸¤ÀÃþÀ³¥Î ¡]2_input 0_hidden _layer
2_ neuron 2_output Perceptron¡^ perceptron222.m
¸ÕµÛ¥Î Matlab ¸Ñ
¡u°Ñ¦Ò®Ñ¥Ø2¡v ®Ñ¤¤Chapter 4 ªº
P4_3¡A¹ï·Ó®Ñ¤¤ªºªÅ¶¡§@¹Ï¡A¥i¥H¹ï Neural Network ¦³§ó¶i¤@¨Bªº¤F¸Ñ¡C
²
§Q¥Î·Pª¾¾¹§@¼Æ¦r¿ëÃÑ
¼Ú§Ó«H
2005.11.15 ¡i°ÝÃD²n»¡©ú¡Bµ{¦¡¡Gper_num_class.m¡j
²
XOR Problem
³o¥i¬O¾ú¥v¤Wªº¤j°ÝÃD³á¡]°Ñ¦Ò®Ñ¥Ø2¦³Â²³æ»¡©ú¡^¡I§A¥i¥H°Ñ¦Ò µ{¦¡ ¡]ºK¦Û °Ñ¦Ò®Ñ¥Ø1¡^¡A½Ð°È¥²¤F¸Ñ¨ÃÅçÃÒ¡I
²
§Q¥ÎHopfield Neural Network§@^¤å¦r¥À¿ëÃÑ ¼Ú§Ó«H
n
´¼¼z«¬¨t²Î±±¨î ½Òµ{¡G
²
¤£±Ñªº¤«¦r¹CÀ¸ ¶ÀªÃ§»
²
LMS Learning and 0~9 Learning ¶ÀªÃ§»
²
©Ð«Î¥ò¤¶¨t²Î S§Ó¨Î
n
°Ñ¦Ò®Ñ¥Ø¡G
1.
¡§Matlab ¤Jªù»P¶i¶¥¡¨¡A»X¥H¥¿ ½sµÛ¡A¾§ªL
2.
¡§Neural Network Desigm,¡¨ Hagan, Demuth,Beale, Thomson Learning,
1996 ¡]¤¤Ä¶¥»¡GÃþ¯«¸gºô¸ô³]p¡A ¨L´f°· ¼f®Õ¡A´¶ªL´µ¹y ¸g¾P°Ó¡^
3.
¡§Ãþ¯«¸gºô¸ô
¡V MATLAB ªºÀ³¥Î¡¨¡AùµØ±j ½sµÛ¡A²M½«¬ì§Þ
4.
Ãþ¯«¸gºô¸ô¤Jªù ¬¡¥ÎMatlab¡A©PÄPµ{ ½sµÛ¡A¥þµØ
5.
Ãþ¯«¸gºô¸ô ²z½×»P¹ê°È¡A±i´´³¹¡B±iÄR¬î¡B¶À¯EÛ µÛ¡AªFµØ®Ñ§½
6.
¾÷¾¹¾Ç²ß¡GÃþ¯«¸gºô¸ô¡B¼Ò½k¨t²Î¥H¤Î°ò¦]ºtºâªk«h¡A§@ªÌ¡GĬ¤ì¬K¡B±i§µ¼w¡A¥þµØ®Ñ§½¡]®Ñ¸¹¡G03324-02¡^
7.
Ãþ¯«¸gºô¸ô»P¼Ò½k±±¨î²z½×¤Jªù¡A§@ªÌ¡G¤ý¶i¼w¡B¿½¤j¥þ¡A¥þµØ®Ñ§½¡]®Ñ¸¹¡G02490-01¡^¡i¤º§tC»y¨¥µ{¦¡¡j