oracle#

Oracle functions for testing model

perceptivo.psychophys.oracle.piecewise_probabilistic(points: numpy.ndarray, scale: float = 5) → callable#

Make a piecewise function along a series of (frequency, amplitude) points with some gaussian error

Parameters

  • points (np.ndarray) – n x 2 array of x/y (frequency, amplitude) coordinates that make up an audiogram

  • scale (float) – Scale parameter of noise in amplitude domain to get answers “wrong”

Returns:

perceptivo.psychophys.oracle.reference_audiogram(scale: float = 2) → callable#

Generate fake audiometry samples using median threshold values obtained from the NHANES dataset: https://wwwn.cdc.gov/Nchs/Nhanes/2015-2016/AUX_I.htm

The median rates make a piecewise linear function:

Frequency

Threshold

500

10

1000

10

2000

10

3000

10

4000

15

6000

20

8000

20
Parameters

scale (float) – amount of randomness to multiply the noise of the pseudo-response threshold by

Returns

callable made by piecewise_probabilistic() that works as an oracle function

Parameters

scale ()

Returns

A numpy piecewise function that returns Sample objects for a given input frequency and amplitude

perceptivo.psychophys.oracle.generate_samples(n_samples: int, scale: float = 2, freqs=None, amplitudes=None, randomize=False, freq_range=(500, 8000), amplitude_range=(0, 50), oracle: Optional[callable] = None) → perceptivo.types.psychophys.Samples#

Generate fake audiometry samples using median threshold values obtained from the NHANES dataset: https://wwwn.cdc.gov/Nchs/Nhanes/2015-2016/AUX_I.htm

The median rates make a piecewise linear function:

Frequency

Threshold

500

10

1000

10

2000

10

3000

10

4000

15

6000

20

8000

20

Examples

from perceptivo.psychophys.oracle import generate_samples

samples = generate_samples(n_samples=1000, scale=10)
samples.plot()

(Source code)

Parameters

  • n_samples (int) – number of samples to generate

  • scale (float) – amount of randomness to multiply the noise of the pseudo-response threshold by

  • freqs (arraylike) – (Optional) - predetermined array of frequencies (of length n_samples) to test

  • amplitudes (arraylike) – (Optional) - predetermined array of amplitudes (of length n_samples) to test

  • randomize (bool) – Randomize order of samples before returning, (default False)

Returns

types.psychophys.Samples