基于特征加权距离的双指数模糊子空间聚类算法 - 控制与决策.pdf

绗?5 鍗 绗? 鏈 鎺 鍒 涓 鍐 绛 2010 骞 8 鏈 Vol. 25 No. 8 Control and Decision Aug. 2010 鏂囩珷缂栧彿: 1001-0920 (2010) 08-1207-04 鍩轰簬鐗瑰緛鍔犳潈璺濈鐨勫弻鎸囨暟妯＄硦瀛愮┖闂磋仛绫荤畻娉 1,2 2 2 鐜 楠 , 鐜嬪＋鍚 , 鐜嬫檽鏄 (1. 鍗椾含鐞嗗伐澶у璁＄畻鏈虹瀛︿笌鎶€鏈闄紝鍗椾含210094 锛?. 姹熷崡澶у淇℃伅宸ョ▼瀛﹂櫌锛屾睙鑻忔棤閿?14122) 鎽 瑕? 浼犵粺鐨勬ā绯婅仛绫荤畻娉?FCM) 浣跨敤娆ф皬璺濈璁＄畻鏁版嵁鐐逛箣闂寸殑宸紓鏃? 瀵逛簬楂樼淮鏁版嵁闆嗚仛绫绘晥鏋滀笉澶熺悊鎯? 瀵规, 浠CM 绠楁硶鐨勭洰鏍囧嚱鏁颁负鍩虹, 鐢ㄧ壒寰佸姞鏉冭窛绂讳唬鏇夸紶缁熺殑娆ф皬璺濈, 鍚屾椂鍚戠害鏉熸潯浠朵腑寮曞叆鎸囨暟饊仧 鍜岎€嫿, 鎻 鍑轰簡涓€绉嶅熀浜庣壒寰佸姞鏉冭窛绂荤殑鍙屾寚鏁版ā绯婂瓙绌洪棿鑱氱被绠楁硶, 骞惰璁轰簡璇ョ畻娉曠殑鏀舵暃鎬? 瀹為獙琛ㄦ槑, 鎵€鎻愬嚭绠楁硶鍙互鏈 鏁堟彁鍙栭珮缁存暟鎹泦鍚勭被鍒殑鐩稿叧鐗瑰緛, 鍦ㄧ湡瀹炴暟鎹泦涓婃湁杈冨ソ鐨勮仛绫绘晥鏋? 鍏抽敭璇岤 妯＄硦鑱氱被锛涚壒寰佸姞鏉冭窛绂伙紱鍏ㄥ眬鏀舵暃鎬э紱闈炲钩琛℃暟鎹泦 涓浘鍒嗙被鍙耳 TP391.4 鏂囩尞鏍囪瘑鐮? A Double-indices fuzzy subspace clustering algorithm based on feature weighted distance 1,2 2 2 WANG Jun , WANG Shi-tong , WANG Xiao-ming (1. School of Computer Science and Technology锛孨anjing University of Science and Technology 锛孨anjing, 210094, China 锛?. School of Information Engineering锛孞iangnan University 锛學uxi 214122 锛孋hina 锛嶤orrespondent 锛歐ANG Shi-tong锛孍-mail 锛歸xwangst@) Abstract: The conventional fuzzy clustering algorithms(FCM) fall short when clustering is performed in high dimensional spaces, because they use the Euclidean metric to compute the distance between data points. In this paper, a fuzzy subspace clustering algorithm is proposed by introducing the feature weighted distance and the power exponent 饊仧 and 饊嫿 into the objective function of FCM. The global convergence property of the proposed algorithm is discussed. The experimental results on real dataset show the effectivenes