def __init__(self, input_shape, graph): with graph.as_default(): self.global_step = tf.Variable(0, name="global_step", trainable=False) self.inp = tf.placeholder(tf.float32, [None, input_shape], name="input") self.out = tf.placeholder(tf.float32, [None], name="output") self.weights = tf.get_variable("weights", [input_shape, 1]) self.bias = tf.get_variable("bias", [1]) self.predictions, self.loss = mean_squared_error_regressor( self.inp, self.out, self.weights, self.bias)
def linear_regression(X, y): """Creates linear regression TensorFlow subgraph. Args: X: tensor or placeholder for input features. y: tensor or placeholder for target. Returns: Predictions and loss tensors. """ with tf.variable_scope('linear_regression'): weights = tf.get_variable('weights', [X.get_shape()[1], 1]) bias = tf.get_variable('bias', [1]) return mean_squared_error_regressor(X, y, weights, bias)
def linear_regression(X, y): """Creates linear regression TensorFlow subgraph. Args: X: tensor or placeholder for input features. y: tensor or placeholder for target. Returns: Predictions and loss tensors. """ with tf.variable_scope('linear_regression'): tf.histogram_summary('linear_regression.X', X) tf.histogram_summary('linear_regression.y', y) y_shape = y.get_shape() if len(y_shape) == 1: output_shape = 1 else: output_shape = y_shape[1] weights = tf.get_variable('weights', [X.get_shape()[1], output_shape]) bias = tf.get_variable('bias', [output_shape]) tf.histogram_summary('linear_regression.weights', weights) tf.histogram_summary('linear_regression.bias', bias) return mean_squared_error_regressor(X, y, weights, bias)