1. The Supervised Network Self- Organizing Map for Classification of Large Data Sets Authors: Papadimitriou et al, Advisor: Dr. Hsu Graduate: Yu-Wei Su

2. Outline Motivation Objective Introduction The Supervised Network SOM The classification partition SOM(CP-SOM) The supervised expert network Applications Conclusions Personal opinion

3. Motivation Real data sets are frequently characterized by the large number of noisy observations Unsupervised learning schemes usually can’t discriminate well over the state space of complex decision boundaries

4. Objective To develop the Network Self-Organizing Map(SNet-SOM) to handle to ambiguous regions of state space To develop more computationally efficient unsupervised learning scheme

5. Introduction SNet-SOM utilizes a two stage learning process that identifying and classifying at the simple regions and supervised learning for the difficult ones The simple regions is handled by the SNet- SOM based on SOM of Kohonen The basic SOM is modified with a dynamic node insertion/deletion process with an entropy-based criterion

6. Introduction( cont.) The difficult regions is handled by supervised learning process such as RBF(radial basis function) or SVM(support vector machine)

7. The Supervised Network SOM The SNet-SOM consists of two components The classification partition SOM(CP-SOM) The supervised expert network

8. CP-SOM The size of the CP-SOM is dynamically expanded with an adaptive process for the ambiguous regions The dynamic growth is based on the entropy- base criterion Classifcation are performed only at the unambiguous part of state space that corresponds to the neurons of small entropy

9. CP-SOM( cont.) CP-SOM learning flow Initialization phase Usually four nodes to represent the input data It has lighter computational demands because of avoiding the fine-tuning of neurons and the small size Adaptation phase However, parameters do not need to shrink with time because the neighborhood is large enough to include the whole and during subsequent training epochs, the neighborhood becomes localized near the winning neuron

10. CP-SOM( cont.) Expansion phase The controlling the number of training patterns that correspond to the ambiguous regions is the motivation for modifying SOM The expansion phase follows the adaptation phase SupervisedExpertMaxPatterns specified a limitation of training set SupervisedExpertMinPatterns specified the lower bound of training set

11. CP-SOM( cont.) 1. To compute the entropy for every node I 2. Detection of the neurons whose are ambiguous according entropy threshold value 3. Evaluation of the map to compute the number of training patterns that correspond to the ambiguous neurons denoted by NumTrainingSetAtAmbiguous

12. CP-SOM( cont.) 4. If NumTrainingSetAtAmbiguous > SupervisedExpertMaxPatterns 1. Perform map expansion by inserting smoothly at the neighborhood of each ambiguous neuron a number of neurons that depends on its fuzziness 2. Repeat the adaptation phase after the dynamic extension else If NumTrainingSetAtAmbiguous < Supervised ExpertMinPatterns Reduce the parameter NodeEntropyThresholdForConsideringAmbiguous and more node will be as ambiguous. Restarting from step 2

13. CP-SOM( cont.) else generate training and testing set for the supervised expert endif The assignment of a class label to each neuron of the CP-SOM is performed by majority-voting scheme As a local averaging operator defined over the class labels of all the patterns that activate neuron as the winner

14. The supervised expert network Has the task of discriminating over the state space regions where are complex class decision boundaries Appropriate neural network models are Radial Basis Function(RBF) and the Support Vector Machines(SVM)

15. RBF supervising expert Obtaining generalization performance by which obtain a tradeoff between the fitness of the solution to the training set and smoothness of the solution The tradeoff cost function as :( positive real number called regularization para., D a stabilizer) where

16. RBF supervising expert( cont.) Proper generalization performance is a difficult issue as well as the selection of centers and para. Supervised ExpertMinPatterns is hard to estimated

17. SVM supervising expert SVM obtains high generalization performance without prioir knowledge even dimension of input space is high The classification is to estimate a function f:R N  {±1} using input-output training data (x1,y1),…(xl,yl) R N x {±1}

18. SVM supervising expert(cont.) To minimized the risk in order to obtain generalization performance Since P(x,y) is unknown,can only minimize the empirical risk

19. SVM supervising expert(cont.) R[f] has the dependence on the VC dim para. h which is done by maximum separation Δ between different classes with linear hyperplane For a set of pattern vector x1,…,xl X, the hyperplanes can be as {x X: w.x+b=0}, w a weight vector, b a bias

20. SVM supervising expert(cont.)

21. xi·w+b≥+1 for yi=+1 (1) yi(xi·w+b)-1≥0,i=1,…,l xi·w+b≤-1 for yi= -1 (2), w is the Normal vector of H1,H2 H1: xi·w+b=1,H2: xi·w+b=-1 Margin=2/║w║, ◎ is a support vector.

22. Applications synthesis data distinction of chaos from noise ischemia detection

23. Synthesis data The synthesis model look like: Construction steps: 1. Generation of some proper value V i,i=1,…,N 2. Induction of observation noise,V’ i =1,…,N 3. Computation of the values of outcome variables 4. Induce observation noise to the outcome variables,O’

24. Synthesis data( cont.)

25. distinction of chaos from noise To design a classification system that is able distinguishing between a three-dim vector and random Gaussian noise Lorenz chaotic system has been used to generate a chaotic trajectory that lying at the three-dim space The difficulty of distinguishing noise is dependent on the state space region

26. distinction of chaos from noise( cont.) The regions far from the attractor can be handle effectively with the CP-SOM classification Rest regions with supervised expert can’t be distinguished since these are regions where the classes overlap Training set :20,000 half of Lorenz system and half of Gaussian noise Test set:20,000 which is constructed similarly

27. distinction of chaos from noise( cont.) Size of ambiguous pattern set was near 2000 with entropy criterion 0.2 Plain SOM avg. performance 79% SNet-SOM with RBF 81% SNet-SOM with SVM 82%

28. ischemia detection The ECG signals of the European ST-T database, which are a set of longterm Holter recording provided by eight countries From the samples composing each beat, a window of 400 millisec. is selected Signal component forms the input to PCA in order to describe most of its content with a few coefficients

29. ischemia detection( cont.) The term dimensionality reduction refers to that 100-dim data vector X is represented with a vector X of 5-dim A wavelet based denoising technique based on Lipschitz regulariation theory is applied Training set:15,000 ST-T segment from 44,000 beats from 6 records, Two class:normal, ischemic Test set: 15 records with 120,000 ECG beats

30. ischemia detection( cont.)

32. Conclusions To obtain significant computational benefits in large scale problems The SNet-SOM is a modular architecture that can be improved along many directions

33. Personal opinion Provide a director of detecting noise within improved SOM It is a nice reference to my research