Chapter 9 Self-Organizing Maps 《神经网络与 及机器学习》 .pptVIP

Chapter 9 Self-Organizing Maps;Figure 9.1 Two self-organized feature maps.;Figure 9.2 Two-dimensional lattice of neurons, illustrated for a three-dimensional input and four-by-four dimensional output (all shown in blue). ;Figure 9.3 Gaussian neighborhood function.;Figure 9.4 Illustration of the relationship between feature map ? and weight vector wi of winning neuron i.;Table 9.1;Figure 9.7 (a) Two-dimensional distribution produced by a linear input–output mapping. (b) Two-dimensional distribution produced by a nonlinear input–output mapping.;Figure 9.8 (a) Distribution of the input data. (b) Initial condition of the two-dimensional lattice. (c) Condition of the lattice at the end of the ordering phase. (d) Condition of the lattice at the end of the convergence phase. The times indicated under maps (b), (c), and (d) represent the numbers of iterations.;Figure 9.9 (a) Distribution of the two-dimensional input data. (b) Initial condition of the one-dimensional lattice. (c) Condition of the one dimensional lattice at the end of the ordering phase. (d) Condition of the lattice at the end of the convergence phase. The times included under maps (b), (c), and (d) represent the numbers of iterations.;Table 9.2;Figure 9.10 Feature map containing labeled neurons with strongest responses to their respective inputs.;Figure 9.11 Semantic map obtained through the use of simulated electrode penetration mapping. The map is divided into three regions, representing birds (white), peaceful species (grey), and hunters (blue).;Figure 9.12 (a) Single-stage vector quantizer with four-dimensional input. (b) Two-stage hierarchical vector quantizer using two-input vector quantizers. (From S.P. Luttrell, 1989a, British Crown copyright.);Figure 9.13 Two-stage encoding–decoding results, using the binary tree shown in blue in Fig. 9.12, for the compression of correlated Gaussian noise input. Correlation coefficient ρ = 0.85. (From S.P. Luttrell, 1989a, British Crown copyrig