public static void inferenceb(Tensor <double> images, int hidden1Units)
{
// ReSharper disable once LocalVariableHidesMember
const int NUM_CLASSES = 10;
// ReSharper disable once LocalVariableHidesMember
const int IMAGE_SIZE = 28;
using (var scope = Tf.NameScope("scopeName"))
{
var weights = Tf.Variable(
Tf.TruncatedNormal(new[] { IMAGE_PIXELS, hidden1Units }, stddev: 1.0 / Math.Sqrt(IMAGE_PIXELS)),
name: "weights");
var biases = Tf.Variable(Tf.ZerosDouble(new[] { hidden1Units }), name: "biases");
var hidden1 = Tf.Nn.Relu(Tf.MatMul(images, weights) + biases);
}
}
/// <summary>
/// Build the MNIST model up to where it may be used for inference.
/// </summary>
/// <param name="images">Images placeholder, from inputs()</param>
/// <param name="hidden1Units">Size of the first hidden layer</param>
/// <param name="hidden2Units">Size of the second hidden layer</param>
/// <returns>softmax_linear: Output tensor with the computed logits</returns>
public static object inference(Tensor <double> images, int hidden1Units, int hidden2Units)
{
// ReSharper disable once LocalVariableHidesMember
const int NUM_CLASSES = 10;
// ReSharper disable once LocalVariableHidesMember
const int IMAGE_SIZE = 28;
// ReSharper disable once LocalVariableHidesMember
const int IMAGE_PIXELS = IMAGE_SIZE * IMAGE_SIZE;
// Hidden 1
Tensor <double> hidden1, hidden2;
using (var scope = Tf.NameScope("hidden1"))
{
var weights = Tf.Variable(
Tf.TruncatedNormal(new[] { IMAGE_PIXELS, hidden1Units },
stddev: 1.0 / Math.Sqrt(IMAGE_PIXELS)),
name: "weights");
var biases = Tf.Variable(Tf.ZerosDouble(new[] { hidden1Units }), name: "biases");
hidden1 = Tf.Nn.Relu(Tf.MatMul(images, weights) + biases);
}
// Hidden 2
using (var scope = Tf.NameScope("hidden2"))
{
var weights = Tf.Variable(
Tf.TruncatedNormal(new[] { hidden1Units, hidden2Units },
stddev: 1.0 / Math.Sqrt(hidden1Units)),
name: "weights");
var biases = Tf.Variable(Tf.ZerosDouble(new[] { hidden2Units }), name: "biases");
hidden2 = Tf.Nn.Relu(Tf.MatMul(hidden1, weights) + biases);
}
// Linear
using (var scope = Tf.NameScope("softmax_linear"))
{
var weights = Tf.Variable(
Tf.TruncatedNormal(new[] { hidden2Units, NUM_CLASSES },
stddev: 1.0 / Math.Sqrt(hidden2Units)), name: "weights");
var biases = Tf.Variable(Tf.ZerosDouble(new[] { NUM_CLASSES }), name: "biases");
var logits = Tf.MatMul(hidden2, weights) + biases;
return(logits);
}
/*
*
*
* """
# Hidden 1
# with tf.name_scope('hidden1'):
# weights = tf.Variable (
# tf.truncated_normal([IMAGE_PIXELS, hidden1_units],
# stddev=1.0 / math.sqrt(float(IMAGE_PIXELS))),
# name='weights')
# biases = tf.Variable(tf.zeros([hidden1_units]),
# name='biases')
# hidden1 = tf.nn.relu(tf.matmul(images, weights) + biases)
#
#
# Hidden 2
# with tf.name_scope('hidden2'):
# weights = tf.Variable(
# tf.truncated_normal([hidden1_units, hidden2_units],
# stddev=1.0 / math.sqrt(float(hidden1_units))),
# name='weights')
# biases = tf.Variable(tf.zeros([hidden2_units]),
# name='biases')
# hidden2 = tf.nn.relu(tf.matmul(hidden1, weights) + biases)
# Linear
# with tf.name_scope('softmax_linear'):
# weights = tf.Variable(
# tf.truncated_normal([hidden2_units, NUM_CLASSES],
# stddev=1.0 / math.sqrt(float(hidden2_units))),
# name='weights')
# biases = tf.Variable(tf.zeros([NUM_CLASSES]),
# name='biases')
# logits = tf.matmul(hidden2, weights) + biases
# return logits
#
*
*/
}