tfp.experimental.mcmc.resample_stratified
Stay organized with collections
Save and categorize content based on your preferences.
Stratified resampler for sequential Monte Carlo.
tfp.experimental.mcmc.resample_stratified(
log_probs, event_size, sample_shape, seed=None, name=None
)
The value returned from this algorithm is similar to sampling with
expanded_sample_shape = tf.concat([[event_size], sample_shape]), axis=-1)
tfd.Categorical(logits=log_probs).sample(expanded_sample_shape)`
but with values sorted along the first axis. It can be considered to be
sampling events made up of a length-event_size vector of draws from
the Categorical distribution. However, although the elements of
this event have the appropriate marginal distribution, they are not
independent of each other. Instead they are drawn using a low variance
stratified sampling method suitable for use with Sequential Monte
Carlo algorithms.
The sortedness is an unintended side effect of the algorithm that is
harmless in the context of simple SMC algorithms.
This function is based on Algorithm #1 in the paper
[Maskell et al. (2006)][1].
Args |
|---|
log_probs
|
A tensor-valued batch of discrete log probability distributions.
|
event_size
|
the dimension of the vector considered a single draw.
|
sample_shape
|
the sample_shape determining the number of draws.
|
seed
|
PRNG seed; see tfp.random.sanitize_seed for details.
Default value: None (i.e. no seed).
|
name
|
Python str name for ops created by this method.
Default value: None (i.e., 'resample_independent').
|
Returns |
|---|
resampled_indices
|
a tensor of samples.
|
References
[1]: S. Maskell, B. Alun-Jones and M. Macleod. A Single Instruction Multiple
Data Particle Filter.
In 2006 IEEE Nonlinear Statistical Signal Processing Workshop.
http://people.ds.cam.ac.uk/fanf2/hermes/doc/antiforgery/stats.pdf
Except as otherwise noted, the content of this page is licensed under the Creative Commons Attribution 4.0 License, and code samples are licensed under the Apache 2.0 License. For details, see the Google Developers Site Policies. Java is a registered trademark of Oracle and/or its affiliates.
Last updated 2023-11-21 UTC.
[null,null,["Last updated 2023-11-21 UTC."],[],[],null,["# tfp.experimental.mcmc.resample_stratified\n\n\u003cbr /\u003e\n\n|------------------------------------------------------------------------------------------------------------------------------------------------------------------|\n| [View source on GitHub](https://github.com/tensorflow/probability/blob/v0.23.0/tensorflow_probability/python/experimental/mcmc/weighted_resampling.py#L317-L379) |\n\nStratified resampler for sequential Monte Carlo. \n\n tfp.experimental.mcmc.resample_stratified(\n log_probs, event_size, sample_shape, seed=None, name=None\n )\n\nThe value returned from this algorithm is similar to sampling with \n\n expanded_sample_shape = tf.concat([[event_size], sample_shape]), axis=-1)\n tfd.Categorical(logits=log_probs).sample(expanded_sample_shape)`\n\nbut with values sorted along the first axis. It can be considered to be\nsampling events made up of a length-`event_size` vector of draws from\nthe `Categorical` distribution. However, although the elements of\nthis event have the appropriate marginal distribution, they are not\nindependent of each other. Instead they are drawn using a low variance\nstratified sampling method suitable for use with Sequential Monte\nCarlo algorithms.\nThe sortedness is an unintended side effect of the algorithm that is\nharmless in the context of simple SMC algorithms.\n\nThis function is based on Algorithm #1 in the paper\n\\[Maskell et al. (2006)\\]\\[1\\].\n\n\u003cbr /\u003e\n\n\u003cbr /\u003e\n\n\u003cbr /\u003e\n\n| Args ---- ||\n|----------------|---------------------------------------------------------------------------------------------------------------------------------|\n| `log_probs` | A tensor-valued batch of discrete log probability distributions. |\n| `event_size` | the dimension of the vector considered a single draw. |\n| `sample_shape` | the `sample_shape` determining the number of draws. |\n| `seed` | PRNG seed; see [`tfp.random.sanitize_seed`](../../../tfp/random/sanitize_seed) for details. Default value: None (i.e. no seed). |\n| `name` | Python `str` name for ops created by this method. Default value: `None` (i.e., `'resample_independent'`). |\n\n\u003cbr /\u003e\n\n\u003cbr /\u003e\n\n\u003cbr /\u003e\n\n\u003cbr /\u003e\n\n| Returns ------- ||\n|---------------------|----------------------|\n| `resampled_indices` | a tensor of samples. |\n\n\u003cbr /\u003e\n\n#### References\n\n\\[1\\]: S. Maskell, B. Alun-Jones and M. Macleod. A Single Instruction Multiple\nData Particle Filter.\nIn 2006 IEEE Nonlinear Statistical Signal Processing Workshop.\n\u003chttp://people.ds.cam.ac.uk/fanf2/hermes/doc/antiforgery/stats.pdf\u003e"]]
View source on GitHub