Tune in to the first Women in ML Symposium this Tuesday, October 19 at 9am PST Register now

KthOrderStatistic

public final class KthOrderStatistic

Computes the Kth order statistic of a data set. The current

implementation uses a binary search requiring exactly 32 passes over the input data. The running time is linear with respect to input size. The median-of-medians algorithm is probably faster, but is difficult to implement efficiently in XLA. The implementation imposes a total ordering on floats. The ordering is consistent with the usual partial order. Positive NaNs are greater than positive infinity. Negative NaNs are less than negative infinity. NaNs with distinct payloads are treated as distinct. Subnormal numbers are preserved (not flushed to zero). Positive infinity is greater than all numbers. Negative infinity is less than all numbers. Positive is greater than negative zero. There are less than k values greater than the kth order statistic. There are at least k values greater than or equal to the Kth order statistic. The semantics are not the same as top_k_unique.

Public Methods

Output<Float>
asOutput()
Returns the symbolic handle of a tensor.
static KthOrderStatistic
create(Scope scope, Operand<Float> input, Long k)
Factory method to create a class wrapping a new KthOrderStatistic operation.
Output<Float>
output()

Inherited Methods

Public Methods

public Output<Float> asOutput ()

Returns the symbolic handle of a tensor.

Inputs to TensorFlow operations are outputs of another TensorFlow operation. This method is used to obtain a symbolic handle that represents the computation of the input.

public static KthOrderStatistic create (Scope scope, Operand<Float> input, Long k)

Factory method to create a class wrapping a new KthOrderStatistic operation.

Parameters
scope current scope
Returns
  • a new instance of KthOrderStatistic

public Output<Float> output ()