theanets.feedforward.Regressor

class theanets.feedforward.Regressor(layers, weighted=False, sparse_input=False)

A regression model attempts to produce a target output.

Regression models are trained by optimizing a (possibly regularized) loss that centers around some measurement of error with respect to the target outputs. This regression model implementation uses the mean squared error.

If we have a labeled dataset containing \(m\) \(d\)-dimensional input samples \(X \in \mathbb{R}^{m \times d}\) and \(m\) \(e\)-dimensional paired target outputs \(Y \in \mathbb{R}^{m \times e}\), then the loss that the Regressor model optimizes with respect to the model parameters \(\theta\) is:

\[\mathcal{L}(X, Y, \theta) = \frac{1}{m} \sum_{i=1}^m \| F_\theta(x_i) - y_i \|_2^2 + R(X, \theta)\]

where \(F_\theta\) is the feedforward function that computes the network output, and \(R\) is a regularization function.

A regression model requires the following inputs at training time:

• x: A two-dimensional array of input data. Each row of x is expected to be one data item. Each column of x holds the measurements of a particular input variable across all data items.
• targets: A two-dimensional array of target output data. Each row of targets is expected to be the target values for a single data item. Each column of targets holds the measurements of a particular output variable across all data items.

The number of rows in x must be equal to the number of rows of targets, but the number of columns in these two arrays may be whatever is required for the inputs and outputs of the problem.

__init__(layers, weighted=False, sparse_input=False)

Methods

error(outputs)

Build a theano expression for computing the network error.

Attributes

num_params	Number of parameters in the entire network model.
params	A list of the learnable theano parameters for this network.

error(outputs)

Build a theano expression for computing the network error.

Parameters:

outputs : dict mapping str to theano expression

A dictionary of all outputs generated by the layers in this network.

Returns:

error : theano expression

A theano expression representing the network error.