From: Joint learning sample similarity and correlation representation for cancer survival prediction
Notations | Explanations |
---|---|
\({\textbf {x}}^{v}\in \mathbb {R}^{d^{v}\times N}\) | Sample set of v-th data type |
\(x_{i}^{v}\in \mathbb {R}^{d^{v}}\) | The i-th sample in v-th data type |
\(f_{v}\) | Fully connection neural network used for feature learning |
\({\textbf {w}}_{f_{v}}^{l}\in \mathbb {R}^{m_{l}\times m_{l-1}}\) | The l-th layer weight matrix of neural network \(f_{v}\) |
\({\textbf {y}}^{v}\in \mathbb {R}^{d\times N}\) | The learned feature representation from \({\textbf {x}}^{v}\) with \(f_{v}\) |
\(\chi ^{v,u}\in \mathbb {R}^{d\times d\times N}\) | Interactive map set between data type v and data type u |
\(\chi _{i}^{v,u}\in \mathbb {R}^{d\times d}\) | Interactive map of i-th sample between data type v and data type u |
\({\textbf {y}}^{v,u}\in \mathbb {R}^{d\times N}\) | The correlation representation of \({\textbf {x}}^{v}\) and \({\textbf {x}}^{u}\) |
\({\textbf {y}}\in \mathbb {R}^{(V(V-1)/2)\times N}\) | Fused correlation representation |
\(P^{m}\) | Normalized weight matrix |
\(S^{m}\) | K nearest similarity matrix |
P | Fused similarity matrix |
\(z_{i}\) | The output of graph convolutional network |