Large Margin Graph Embedding-Based Discriminant Dimensionality Reduction

<div>Illustration of embedding space and hypothesis margins: (a) Toy data (H(x1) = -0.83)); (b) MFA (k1 = k2 = 5, H(x1) = -0.84); (c) MNMDP ((k) = 5, H(x1) = 0.19); (d) DNE ((k) = 5, H(x1) = 0.43); (e) DAG-DNE ((k) = 5, H(x1) = -0.84); (f) LMGE-DDR ((k) = k1 = k2 = 3, (a) = 0.8, H(x1) = 0.39).</div>

Scientific Programming

fig4

Figure 4

Figure 4: Large Margin Graph Embedding-Based Discriminant Dimensionality Reduction 