Table 1 Notations.
Variables | Definition |
|---|---|
\(t\) | Time step |
\(P + 1\) | Number of input layer |
\(N\) | Number of \(\sum_{1}\) layer |
\(Q\) | Number of \(\prod\) layer |
\(I(t)\) | overall input at current time step \(t\) |
\(I_{p} (t)\) | The p-th element of \(I(t)\) |
\(f( \cdot )\) | Activation function |
\(y(t - 1)\) | Output value of network at previous time step \(t - 1\) |
\(W_{0}\) | Weight vector between \(\sum_{2}\) layer and \(\prod\) layer |
\(Q\) | Number of nodes of the \(\prod\) layer |
\(w_{0q}\) | The q-th element of \(W_{0}\) |
\(W_{n}\) | The n-th weigh vector of \(\sum_{1}\) layer |
\(\varepsilon (t)\) | Variable of \(\sum_{1}\) layer at current time step \(t\) |
\(\varepsilon_{n} (t)\) | The n-th element of \(\varepsilon (t)\) |
\(A_{q}\) | Set of neurons about the \(\sum_{1}\) layer linked with the \(q - th\) neuron of the \(\prod\) layer |
\(B_{n}\) | Set of neurons about \(\prod\) layer linked with the \(n - th\) neuron of \(\sum_{1}\) layer |
\(a\) | Arbitrary |
\(\phi (a)\) | Number of elements in \(a\) |
\(\delta (t)\) | Output result of \(\prod\) layer at current time step \(t\) |
\(\delta_{q} (t)\) | The q-th element of \(\delta (t)\) |
\(y(t)\) | Actual output of RSPSNN at current time step \(t\) |
