precise-lite-amd64aarch64/precise_lite/threshold_decoder.py
2021-08-15 22:14:36 +01:00

60 lines
2.5 KiB
Python

# Copyright 2019 Mycroft AI Inc.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import numpy as np
from typing import Tuple
from precise_lite.functions import asigmoid, sigmoid, pdf
class ThresholdDecoder:
"""
Decode raw network output into a relatively linear threshold using
This works by estimating the logit normal distribution of network
activations using a series of averages and standard deviations to
calculate a cumulative probability distribution
"""
def __init__(self, mu_stds: Tuple[Tuple[float, float]], center=0.5, resolution=200, min_z=-4, max_z=4):
self.min_out = int(min(mu + min_z * std for mu, std in mu_stds))
self.max_out = int(max(mu + max_z * std for mu, std in mu_stds))
self.out_range = self.max_out - self.min_out
self.cd = np.cumsum(self._calc_pd(mu_stds, resolution))
self.center = center
def decode(self, raw_output: float) -> float:
if raw_output == 1.0 or raw_output == 0.0:
return raw_output
if self.out_range == 0:
cp = int(raw_output > self.min_out)
else:
ratio = (asigmoid(raw_output) - self.min_out) / self.out_range
ratio = min(max(ratio, 0.0), 1.0)
cp = self.cd[int(ratio * (len(self.cd) - 1) + 0.5)]
if cp < self.center:
return 0.5 * cp / self.center
else:
return 0.5 + 0.5 * (cp - self.center) / (1 - self.center)
def encode(self, threshold: float) -> float:
threshold = 0.5 * threshold / self.center
if threshold < 0.5:
cp = threshold * self.center * 2
else:
cp = (threshold - 0.5) * 2 * (1 - self.center) + self.center
ratio = np.searchsorted(self.cd, cp) / len(self.cd)
return sigmoid(self.min_out + self.out_range * ratio)
def _calc_pd(self, mu_stds, resolution):
points = np.linspace(self.min_out, self.max_out, resolution * self.out_range)
return np.sum([pdf(points, mu, std) for mu, std in mu_stds], axis=0) / (resolution * len(mu_stds))