The NeuralNet function displays a message after finishing.
Property | Value |
---|---|
Reached Threshold | The error threshold reached when training stops. |
Threshold Steps | Max iterations reached or convergence reached. |
The query below displays the OutputTable containing the fitted weights:
SELECT * FROM output_table;
This output table has the same schema as the weight table and has a row for each neural network with one column for each weight. As the following table shows, the weight columns are labeled 0, 1, 2, and so on, and are ordered by layer, origin node, and destination node. For example, the weight in column 0 corresponds to w101 in Introduction to Neural Networks, column 1 corresponds to w102, and so on, up to column 12 corresponding to w231.
Initial layer | From neuron | To neuron | Weight | Corresponding column in <OutputTable> |
---|---|---|---|---|
Input layer | X0 | Φ1 | W101 | 0 |
Input layer | X0 | Φ2 | W102 | 1 |
Input layer | X0 | Φ3 | W103 | 2 |
Input layer | X1 | Φ1 | W111 | 3 |
... | ... | ... | ... | ... |
Input layer | X2 | Φ3 | W123 | 8 |
Hidden layer | Φ0 | Output | W201 | 9 |
... | ... | ... | ... | ... |
Hidden layer | Φ3 | Output | W231 | 12 |