From: Testing coverage criteria for optimized deep belief network with search and rescue
Symbols | Descriptions |
---|---|
\(g_{h}\) | The amount of hidden units |
\(g_{v}\) | The amount of visible units |
\(W\) | Simultaneous weights among hidden and visible units |
\(L\) | Simultaneous weights among visible and visible units |
\(J\) | Simultaneous weights among hidden and hidden units |
\(v\) | Visible unit |
\(h\) | Hidden unit |
\(W_{ij}\) | Symmetric connection among hidden (\(j\)) and visible (\(i\)) unit |
\(b_{j}\) and \(a_{i}\) | Bias terms |
\(Z\) | Normalization constant |
\(m\) | Amount of training data samples |
\(X_{il}\) | \(i\;th\) unit of the \(l\;th\) data occasion |
\(\varepsilon\) | Learning rate |
\(g(x) = 1/(1 + \exp ( - x))\) | Logistic sigmoid function |
\(D\) | Entire amount of data samples |
\(O_{z}^{e}\) | Expected output |
\(Z_{z}^{e}\) | Predicted output |
\(M_{n}\) | Location of the \(n\;th\) stored clue |
\(N = \left\{ {n_{1} ,\,n_{2} ,\,...} \right\}\) | Set of neurons |
\(T = \left\{ {x_{1} ,\,x_{2} ,\,...} \right\}\) | Test inputs |
\(L_{i}\) | Arrangement of neurons on the \(i^{th}\) layer |
\(X_{i}\) | Preceding location |
\(T\) | Chosen features |
\(X\) | Weight |
\(Low_{n}\) | Lower boundary output values for a neuron \(n\) |
\(High_{n}\) | Upper boundary output values for a neuron \(n\) |