Send the link below via email or IMCopy
Present to your audienceStart remote presentation
- Invited audience members will follow you as you navigate and present
- People invited to a presentation do not need a Prezi account
- This link expires 10 minutes after you close the presentation
- A maximum of 30 users can follow your presentation
- Learn more about this feature in our knowledge base article
Do you really want to delete this prezi?
Neither you, nor the coeditors you shared it with will be able to recover it again.
Make your likes visible on Facebook?
Connect your Facebook account to Prezi and let your likes appear on your timeline.
You can change this under Settings & Account at any time.
Ensemble of neuro-fuzzy Kohonen networks for adaptive fuzzy clustering
Transcript of Ensemble of neuro-fuzzy Kohonen networks for adaptive fuzzy clustering
Observations applied one-by-one Difficulties The result – segmentation of the original data set into m classes with some level of membership of k-th feature vector to j-th cluster. Initial Information – sample of observations, formed from N n-dimensional feature vectors: Data may contain outliers and noise Clusters can appear and disappear Is this strange observation outlier or new cluster? Clusters can migrate The same observation in different time
may belong to different cluster. Ensemble of adaptive neuro-fuzzy Kohonen networks where Why is not good idea? Ensembles are bad online Ensembles are bad in clustering The task of clustering is incorrect itself,
it's hard to evaluate results of individual networks Generalizing system usually needs batch of data to evaluate results of individual networks Adaptive learning algorithms on each level Self-orginizing mechanism on each level Adaptive possibilistic self-learning algorithm Suppression Robust self-learning algorithm The Ensemble The Ensemble of neuro-fuzzy Kohonen networks for adaptive clustering Bogdan Kolchygin,
3rd year postgraduate student firstname.lastname@example.org Kharkov national university
of radio electronics Control System Research Laboratory
Head: Yevgeniy Bodyanskiy collision Remarks Perspectives Bibliography Non-spherical clusters Quality criteria Adaptive possibilistic clustering algorithm Suppression Adaptive robust fuzzy clustering algorithm Fuzzy Kohonen networks Tsao E. C.-K., Bezdek J. C., Pal N. R. Fuzzy Kohonen clustering networks. –
Pattern Recognition, Volume 27, Issue 5, 1994 – p. 757–764 Bodyanskiy Ye., Kolchygin B., Pliss I. Adaptive neuro-fuzzy Kohonen network
with variable fuzzifier. – International Journal ”Information Theories and Ap-
plications”, 2011. – Vol.18. – #3 – p. 215–223. Fan J.-L., Zhen W.-Z., Xie W.-X. Suppressed Fuzzy C-Means Clustering Algo-
rithm. – Pattern Recognition Letters. – 2003. – 23. – p. 1607–1612. Bodyanskiy Ye, Kolchygin B. Adaptive robust self-learning algorithm for fuzzy
clustering. – Proc. Int. Conf. on Intellectual systems for decision making and
problems of computational intelligence. – Kherson: KNTU, 2012. – p. 367-369
(in russian). The most sophisticated and data-dependent parameter is λ. Individual networks with extreme settings work unstable, but such networks as part of ensemble improve results. Quantity of networks surprisingly hasn't significant effect on results... in most cases. Adjustment algorithm for λ. Gustafson-Kessel? Density clustering? Type-2 Fuzzy System? Over 50 methods, 0 for online systems. There is no analytical solution.