ApplyFtrl

tensorflow C++ API

tensorflow::ops::ApplyFtrl

Update '*var' according to the Ftrl-proximal scheme.

Summary

accum_new = accum + grad * grad linear += grad + (accum_new^(-lr_power) - accum^(-lr_power)) / lr * var quadratic = 1.0 / (accum_new^(lr_power) * lr) + 2 * l2 var = (sign(linear) * l1 - linear) / quadratic if |linear| > l1 else 0.0 accum = accum_new

Arguments:

• scope: A Scope object
• var: Should be from a Variable().
• accum: Should be from a Variable().
• linear: Should be from a Variable().
• grad: The gradient.
• lr: Scaling factor. Must be a scalar.
• l1: L1 regulariation. Must be a scalar.
• l2: L2 regulariation. Must be a scalar.
• lr_power: Scaling factor. Must be a scalar.

Optional attributes (seeAttrs):

• use_locking: If True, updating of the var and accum tensors will be protected by a lock; otherwise the behavior is undefined, but may exhibit less contention.

Returns:

• Output: Same as "var".

ApplyFtrl block

Source link : https://github.com/EXPNUNI/enuSpaceTensorflow/blob/master/enuSpaceTensorflow/tf_training.cpp

Argument:

• Scope scope : A Scope object (A scope is generated automatically each page. A scope is not connected.)
• Input var: connect Input node.
• Input accum: connect Input node.
• Input linear: connect Input node.
• Input grad: connect Input node.
• Input lr: connect Input node.
• Input l1: connect Input node.
• Input l2: connect Input node.
• Input lr_power: connect Input node.
• ApplyFtrl ::Attrs attrs : Input attrs in value. ex) use_locking_ = false;

Return:

• Output output : Output object of ApplyFtrl class object.

Result:

• std::vector(Tensor) result_output : Returned object of executed result by calling session.

Using Method