M. Barni, V. Cappellini, and A. Mecocci, "Comments on `A possibilistic approach to clustering'," IEEE Trans. Fuzzy Syst., vol. 4, pp. 393--396, 1996.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....as possible, while the second term trades memberships as large as possible in order to avoid the trivial solution. As it is converged to a local or global minimum, a near optimal or optimal solution should be obtained. However, when the PCM algorithm is applied to test, the coincidental cluster [6] is also obtained both on the simulated data and on the real image tests. In Fig. 3(b) the two clusters are merged into just one cluster by the PCM algorithm. To solve this problem, Krishnapuram et al. 7] modified the membership function equation as follows, u d ij ij = exp( h . 2) ....

M. Barni, V. Cappellini and A. Mecocci, "Comments on `A possibilistic approach to clustering'," IEEE Trans. Fuzzy Syst., Vol. 4, No. 3, pp. 393-396, 1996.

....process in traditional clustering algorithms. If u ik is the the largest, then its weighting, w ik should be the largest among w ik for all k. Through this kind of normalization like process, we let each cluster prototype interact with each other to prevent generating coincident clusters [14]. Step 2) Updating Cluster Fuzzy Prototypes: We update the cluster prototype by m 0 i = P n k=1 w 2 ik x k P k=n k=1 w 2 ik # (4) oe 0 i = P n k=1 w ik fl fl flx i ;m i fl fl fl P k=n k=1 w ik # (5) where, m 0 i is the new cluster centroid of i th cluster and ....

M. Barni, V. Cappellini, and A. Mecocci, "Comments on a possibilistic approachto clustering," in IEEE Trans. Fuzzy System, vol. 4, pp. 393--396, 1996.

....that are possibilistic or absolute (i.e. not relative) 15] 16] This leads to systems, e.g. the Possibilistic c means clustering algorithm [15] where cluster centers are sought independently of one another. These systems are affected by the tendency to produce coincident clusters [16] [17]. Notice that Eq. 2) can also be interpreted as a link between clustering problems and distance weighted regression [13] see Appendix 1) 2 Optimization issues The necessary condition that guarantees approximate minimization of Eq. 2) is E gen C j = Gamma m X i=1 (X i Gamma C j )k j ....

M. Barni, V. Cappellini, and A. Mecocci, "Comments on a possibilistic approach to clustering," IEEE Trans. Fuzzy Systems, vol. 4, no. 3, pp. 393-396, 1996.

.... may have significantly high probabilistic membership values and may severely affect the prototype parameter estimate (e.g. refer to [62] On the other hand, in possibilistic fuzzy clustering, learning rates computed from absolute typicalities tend to produce coincident clusters [62] [63]. This poor behavior can be explained by the fact that cluster prototypes are uncoupled in possibilistic clustering, i.e. possibilistic clustering algorithms try to minimize an objective function by operating on each cluster independently. This leads to an increase in the number of local minima. ....

M. Barni, V. Cappellini, and A. Mecocci, "Comments on a possibilistic approach to clustering," IEEE Trans. Fuzzy Systems, vol. 4, no. 3, pp. 393-396, 1996.

No context found.

M. Barni, V. Cappellini, and A. Mecocci, "Comments on `A possibilistic approach to clustering'," IEEE Trans. Fuzzy Syst., vol. 4, pp. 393--396, 1996.

No context found.

M. Barni, V. Cappellini, and A. Mecocci, "Comments on a possibilistic approach to clustering", in IEEE Transactions Fuzzy System,volume4, pages 393--396, 1996.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC