เอกสารการประชุมวิชาการและเสนอผลงานวิจัย มหาวิทยาลัยทักษิณ ครั้งที่ 19 2552 - page 208

„µ¦ž¦³»¤ª·µ„µ¦Â¨³Áœ°Ÿ¨Šµœª·‹´¥ ¤®µª·š¥µ¨´¥š´„¬·– ‡¦´ÊŠš¸É 19 ž¦³‹Îµže 2552
4
‹³Á®È œÅ—oªn µ¤¸ „µ¦š—¨°Š 2 ‡¦´ÊŠ ¨³ o °¤¼ ¨š¸Éčo ĜäÁ—¨Á‡¦º ° nµ¥Ä¥ž¦³µšÁš¸ ¥¤œ´Êœ™¼ „n Š
°°„Áž}œ 2 » — ‡º° » — o°¤¼ ¨š¸Éčo Ĝ„µ¦ f„°œ ¨³» — °Š„µ¦š—° Ž¹É Š»— o°¤¼¨š¸ÉčoĜ„µ¦š—¨°Š
œ´ ʜ ‡º° œ´ „«¹„¬µže » —šoµ¥ °Š»— o°¤¼¨ ®¦º°œ´„«¹„¬µš¸É ¤¸¦®´ œ´ „«¹„¬µ ¹Êœ˜oœ—oª¥ 50 čo Î µ®¦´„µ¦š—¨°Šš¸É 1
nŠ°°„Áž}œ »— o°¤¼¨š¸ Éčo Ĝ„µ¦ f„°œ 72.65% ¨³ »— °Š„µ¦š—° 27.35% 宦´„µ¦š—¨°Š
‡¦´Ê Šš¸É 2 ŗo‹Îµœªœ o°¤¼¨ °Šœ´„«¹„¬µš¸É š¦µŸ¨„µ¦Á„·—ž{ ®µ Á oµÅž ‹µ„Á—·¤ ¤¸ ‹Îµœªœ 245 ¦µ¥„µ¦ Áž}œ 260
¦µ¥„µ¦ ¨³Å—onŠ o°¤¼¨ °°„Áž} œ »— o°¤¼¨š¸Éčo Ĝ„µ¦ f„°œ 68.46% ¨³ » — °Š„µ¦š—° 31.54%
„µ¦„ε®œ—˜´ ªÂž¦ ¨³„µ¦Á oµ¦®´ ĜŠµœª·‹´¥œ¸Ê Ÿ¼oª·‹´¥Å—o¡· ‹µ¦–µ‹µ„ o°¤¼¨œ´ „«¹„¬µš¸É™¼„Á„ȝĜ
¢j ¤°µ‹µ¦¥rš¸É ž¦¹„¬µ Ž¹ÉŠž¦³„°Åž—oª¥ o°¤¼¨Äœnªœ °Šž¦³ª´˜· n ªœ˜´ª °Šœ´„«¹ „¬µ ž¦³„°—o ª¥ o°¤¼¨
œ´ „«¹„¬µÁ°Š o °¤¼¨„µ¦«¹ „¬µ‹µ„™µ´ œÁ—·¤ ¨³ o °¤¼¨Ÿ¼ož„‡¦°Š °Šœ´ „«¹„¬µ Ž¹ÉŠÄo Áž} œ˜´ ªÂž¦Äœ„µ¦
ž¦³¤ª¨Ÿ¨Ã®œ—„µ¦¦´Ÿ¨ ¤¸š´Ê Š·Êœ 17  œ°„‹µ„œ¸Ê Ÿ¼oª·‹´¥Å—o čo®´ª o°Äœ„µ¦‡´ —„¦°Šœ´„«¹„¬µÁž}œ o°¤¼¨
ž¦³„° ¨³Áž}œ˜´ ªÂž¦š¸Éčo 宦´„µ¦¦´ o°¤¼¨ (Input Node) ×¥„µ¦ª· Á‡¦µ³®r°Š„µ¦‡´ —„¦°Šœ´„«¹„¬µ
œ´Ê œŸ¼oª·‹´¥Å—o œª·›¸ „µ¦‡·—‡nµ‡³ÂœœÎµ®¦´„µ¦ª·Á‡¦µ³®r ªnµœ´„«¹„¬µ Áž}œ»‡‡¨š¸Éž„˜· ®¦º°Á¸É¥Š/¤¸ž{®µ ¤¸‹Îµœªœ
š´ÊŠ·Êœ 23  ¦» žÅ—oªnµ o °¤¼¨œ´„«¹ „¬µš¸ÉÁž}œ˜´ªÂž¦Á„¸É ¥ª„´ž¦³ª´ ˜· nªœ˜´ ª o°¤¼¨œ´„«¹ „¬µš¸ÉÁž}œ
˜´ªÂž¦Á„¸É¥ª„´„µ¦‡´—„¦°Šœ´„«¹„¬µ ‹³Á®È œªnµ„µ¦Â—ŠÃ®œ— °Š˜´ªÂž¦œ· ª¦°¨ÄœŠµœª·‹´¥œ¸Ê ¦»žÅ—oªnµŸ¼oª·‹´¥
ŗo „ε®œ—˜´ªÂž¦ 40 
„µ¦Å—o¤µ °Š¦¼žÂ ®¦º°Ã¤Á—¨Îµ®¦´„µ¦¡¥µ„¦–r ð„µ„µ¦Á„·—ž{®µ °Šœ´„«¹ „¬µ nŠ
°°„Áž} œ 3 nªœ ž¦³„°—oª¥ ´Êœ¦´ o°¤¼¨ (Input Layer) ŗo„ε®œ—‹ÎµœªœÃ®œ—Äœ´Êœ¦´ o°¤¼¨Ášnµ„´‹Îµœªœ
˜´ªÂž¦š´ÊŠ®¤— °Š»— o°¤¼¨š¸É¦´¤µ ‹Îµœªœ 40  ´ÊœŽn°œÁ¦oœ (Hidden Layer) „µ¦„ε®œ—Žn °œÁ¦oœœ´Êœ
Ÿ¼oª·‹´¥Å—o„ε®œ—Áž}œ¡µ¦µ¤·Á˜°¦r (Parameter) š¸É ε‡´°¥nµŠ®œ¹ÉŠš¸ÉčoĜ„µ¦Äo ĜäÁ—¨ —´Šœ´ÊœÄœ„µ¦ª·‹´¥Äœ‡¦´ÊŠœ¸Ê
‹¹ ŠÅ—o¡·‹µ¦–µÁ¨º °„čo Ĝ´ÊœŽn °œÁ¦oœ‹µ„„µ¦š—¨°Š®¨µ¥ Ç ‡¦´Ê Š ‹µ„‹ÎµœªœÃ®œ—š¸É ˜„˜n µŠ„´œ ´Êœ
—ŠŸ¨ (Output Node) Ĝ„µ¦ª·‹´¥œ¸Ê Ÿ¼oª·‹´ ¥Å—ošÎ µ„µ¦„ε®œ—‹ÎµœªœÃ®œ—„µ¦Â—ŠŸ¨Äœ„µ¦š—¨°Š‡¦´Ê Šš¸É 1
‹Îµœªœ 1  ¨³‹ÎµœªœÃ®œ—Äœ„µ¦Â—ŠŸ¨Äœ„µ¦š—¨°Š‡¦´ÊŠš¸É 2 ‹Î µœªœ 2  äÁ—¨
Back – Propagation Neuron Network 宦´¡¥µ„¦–rð„µÁ„·—ž{®µ °Šœ´„«¹ „¬µ ‹µ„„µ¦š—¨°Š‡¦´ÊŠš¸É 1 ¤¸
‹Îµœªœ „µ¦¦´ o°¤¼¨Ášnµ„´‹Îµœªœ˜´ªÂž¦ ‹Îµœªœ 40  Ž¹ÉŠŸ¼oª·‹´¥¡·‹µ¦–µ‹µ„¢j¤ o°¤¼¨Šµœš³Á¸¥œ
‹µ„„°ŠnŠÁ¦· ¤ª·µ„µ¦Â¨³Â¢j ¤°µ‹µ¦¥rš¸Éž¦¹ „¬µ ¨³¤¸‹ÎµœªœÃ®œ—„µ¦Â—ŠŸ¨ ‹Îµœªœ 1  Ášn µœ´Êœ
äÁ—¨ Back – Propagation Neuron Network 宦´¡¥µ„¦–rð„µÁ„·—ž{®µ °Šœ´„«¹„¬µ ‹µ„„µ¦š—¨°Š‡¦´Ê Šš¸É
2 ¤¸‹Îµœªœ „µ¦¦´  o°¤¼¨Ášnµ„´‹Îµœªœ˜´ªÂž¦ ‹Îµœªœ 40  ¨³¤¸‹ÎµœªœÃ®œ—„µ¦Â—ŠŸ¨ ‹Îµœªœ
2 
„µ¦ª· Á‡¦µ³®ro°¤¼¨š¸ÉŸ¼oª·‹´¥Å—očo 宦´ „µ¦š—¨°Šœ´ÊœÅ—o Á˜¦¸¥¤ o°¤¼¨Î µ®¦´„µ¦Áš¦œœ·ª¦°¨Ã—¥
µ¤µ¦™¦³» ‡nµ˜nµŠ Ç š¸ÉÁ„¸É¥ª o°Š„´„µ¦Áš¦œÅ—o Án œ ‡nµ„µ¦Á¦¸¥œ¦¼o (Epoch), Hidden Node, ‡nµ‡»–¤´˜·š¸É‹³Äo
Áš¦œ ®¨´Š‹µ„œ´ÊœÃž¦Â„¦¤‹³šÎµ„µ¦Áš¦œœ· ª¦°¨Ã—¥ Back - Propagation Training Algorithm ×¥„µ¦ž¦´‡nµ
Weight ¨³ Bias Ĝš»„ Ç 1 ¦° ¨³‹³ž¦´„¦³š´É Š Training Output ¨³ Target Output ¤¸‡nµÄ„¨oÁ‡¸¥Š„´œ
o°¤¼ ¨°·œ¡»š (Input) ®¦º°´Ê œ¦´ o °¤¼¨ Ÿ¼oª·‹´ ¥Å—oÁ˜¦¸¥¤Åªo ‹Îµœªœ 40  ¨³ 35  „µ¦„ε®œ—
‡nµ¡µ¦µ¤·Á˜°¦r ˜nµŠ Ç š¸Éčo Î µ®¦´„µ¦Áš¦œ ‡º° Ĝ´ÊœŽn°œÁ¦oœ (Hidden Node) ‹ÎµœªœÃ®œ—Äœ´Ê œŽn°œÁ¦o œ
Áž}œ¡µ¦µ¤·Á˜°¦r š¸É ε‡´°¥n µŠ®œ¹ÉŠÄœÃ¤Á—¨Ä¥ž¦³µšÁš¸¥¤ —´Šœ´Êœ‹¹ŠÅ—o¡· ‹µ¦–µÁ¨º°„čo‹µ„„µ¦š—¨°Š