/
negative_binomial.py
197 lines (169 loc) · 7.54 KB
/
negative_binomial.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
# Copyright 2017 The TensorFlow Authors. All Rights Reserved.
#
# 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.
# ==============================================================================
"""The Negative Binomial distribution class."""
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
from tensorflow.python.framework import dtypes
from tensorflow.python.framework import ops
from tensorflow.python.framework import tensor_shape
from tensorflow.python.ops import array_ops
from tensorflow.python.ops import check_ops
from tensorflow.python.ops import math_ops
from tensorflow.python.ops import random_ops
from tensorflow.python.ops.distributions import distribution
from tensorflow.python.ops.distributions import util as distribution_util
class NegativeBinomial(distribution.Distribution):
"""NegativeBinomial distribution.
The NegativeBinomial distribution is related to the experiment of performing
Bernoulli trials in sequence. Given a Bernoulli trial with probability `p` of
success, the NegativeBinomial distribution represents the distribution over
the number of successes `s` that occur until we observe `f` failures.
The probability mass function (pmf) is,
```none
pmf(s; f, p) = p**s (1 - p)**f / Z
Z = s! (f - 1)! / (s + f - 1)!
```
where:
* `total_count = f`,
* `probs = p`,
* `Z` is the normalizaing constant, and,
* `n!` is the factorial of `n`.
"""
def __init__(self,
total_count,
logits=None,
probs=None,
validate_args=False,
allow_nan_stats=True,
name="NegativeBinomial"):
"""Construct NegativeBinomial distributions.
Args:
total_count: Non-negative floating-point `Tensor` with shape
broadcastable to `[B1,..., Bb]` with `b >= 0` and the same dtype as
`probs` or `logits`. Defines this as a batch of `N1 x ... x Nm`
different Negative Binomial distributions. In practice, this represents
the number of negative Bernoulli trials to stop at (the `total_count`
of failures), but this is still a valid distribution when
`total_count` is a non-integer.
logits: Floating-point `Tensor` with shape broadcastable to
`[B1, ..., Bb]` where `b >= 0` indicates the number of batch dimensions.
Each entry represents logits for the probability of success for
independent Negative Binomial distributions and must be in the open
interval `(-inf, inf)`. Only one of `logits` or `probs` should be
specified.
probs: Positive floating-point `Tensor` with shape broadcastable to
`[B1, ..., Bb]` where `b >= 0` indicates the number of batch dimensions.
Each entry represents the probability of success for independent
Negative Binomial distributions and must be in the open interval
`(0, 1)`. Only one of `logits` or `probs` should be specified.
validate_args: Python `bool`, default `False`. When `True` distribution
parameters are checked for validity despite possibly degrading runtime
performance. When `False` invalid inputs may silently render incorrect
outputs.
allow_nan_stats: Python `bool`, default `True`. When `True`, statistics
(e.g., mean, mode, variance) use the value "`NaN`" to indicate the
result is undefined. When `False`, an exception is raised if one or
more of the statistic's batch members are undefined.
name: Python `str` name prefixed to Ops created by this class.
"""
parameters = locals()
with ops.name_scope(name, values=[total_count, logits, probs]):
self._logits, self._probs = distribution_util.get_logits_and_probs(
logits, probs, validate_args=validate_args, name=name)
with ops.control_dependencies(
[check_ops.assert_positive(total_count)] if validate_args else []):
self._total_count = array_ops.identity(total_count)
super(NegativeBinomial, self).__init__(
dtype=self._probs.dtype,
reparameterization_type=distribution.NOT_REPARAMETERIZED,
validate_args=validate_args,
allow_nan_stats=allow_nan_stats,
parameters=parameters,
graph_parents=[self._total_count, self._probs, self._logits],
name=name)
@property
def total_count(self):
"""Number of negative trials."""
return self._total_count
@property
def logits(self):
"""Log-odds of a `1` outcome (vs `0`)."""
return self._logits
@property
def probs(self):
"""Probability of a `1` outcome (vs `0`)."""
return self._probs
def _batch_shape_tensor(self):
return array_ops.broadcast_dynamic_shape(
array_ops.shape(self.total_count),
array_ops.shape(self.probs))
def _batch_shape(self):
return array_ops.broadcast_static_shape(
self.total_count.get_shape(),
self.probs.get_shape())
def _event_shape_tensor(self):
return array_ops.constant([], dtype=dtypes.int32)
def _event_shape(self):
return tensor_shape.scalar()
def _sample_n(self, n, seed=None):
# Here we use the fact that if:
# lam ~ Gamma(concentration=total_count, rate=(1-probs)/probs)
# then X ~ Poisson(lam) is Negative Binomially distributed.
rate = random_ops.random_gamma(
shape=[n],
alpha=self.total_count,
beta=math_ops.exp(-self.logits),
dtype=self.dtype,
seed=seed)
return random_ops.random_poisson(
rate,
shape=[],
dtype=self.dtype,
seed=distribution_util.gen_new_seed(seed, "negative_binom"))
def _cdf(self, positive_counts):
if self.validate_args:
positive_counts = math_ops.floor(
distribution_util.embed_check_nonnegative_discrete(
positive_counts, check_integer=False))
return math_ops.betainc(
self.total_count, positive_counts + 1.,
math_ops.sigmoid(-self.logits))
def _log_prob(self, positive_counts):
return (self._log_unnormalized_prob(positive_counts)
- self._log_normalization(positive_counts))
def _log_unnormalized_prob(self, positive_counts):
if self.validate_args:
positive_counts = distribution_util.embed_check_nonnegative_discrete(
positive_counts, check_integer=True)
return self.total_count * math_ops.log1p(
-self.probs) + positive_counts * math_ops.log(self.probs)
def _log_normalization(self, positive_counts):
if self.validate_args:
positive_counts = distribution_util.embed_check_nonnegative_discrete(
positive_counts, check_integer=True)
return (-math_ops.lgamma(self.total_count + positive_counts)
+ math_ops.lgamma(positive_counts + 1.)
+ math_ops.lgamma(self.total_count))
def _mean(self):
return self.total_count * math_ops.exp(self.logits)
def _mode(self):
adjusted_count = array_ops.where(
1. < self.total_count,
self.total_count - 1.,
array_ops.zeros_like(self.total_count))
return math_ops.floor(adjusted_count * math_ops.exp(self.logits))
def _variance(self):
return self._mean() / math_ops.sigmoid(-self.logits)