Measuring Sample Discrepancy with Diffusions
38 mins 54 secs,
156.23 MB,
WebM
640x360,
29.97 fps,
44100 Hz,
548.36 kbits/sec
Share this media item:
Embed this media item:
Embed this media item:
About this item
Description: |
Duncan, A
Tuesday 18th July 2017 - 15:40 to 16:20 |
---|
Created: | 2017-07-19 13:03 |
---|---|
Collection: | Scalable inference; statistical, algorithmic, computational aspects |
Publisher: | Isaac Newton Institute |
Copyright: | Duncan, A |
Language: | eng (English) |
Distribution: | World (downloadable) |
Explicit content: | No |
Aspect Ratio: | 16:9 |
Screencast: | No |
Bumper: | UCS Default |
Trailer: | UCS Default |
Abstract: | In many applications one often wishes to quantify the discrepancy between a sample and a target probability distribution. This has become particularly relevant for Markov Chain Monte Carlo methods, where practitioners are now turning to biased methods which trade off asymptotic exactness for computational speed. While a reduction in variance due to more rapid sampling can outweigh the bias introduced, the inexactness creates new challenges for parameter selection. The natural metric in which to quantify this discrepancy is the Wasserstein or Kantorovich metric. However, the computational difficulties in computing this quantity has typically dissuaded practitioners. To address this, we introduce a new computable quality measure based on Stein's method that quantifies the maximum discrepancy between sample and target expectations over a large class of test functions. We demonstrate this tool by comparing exact, biased, and deterministic sample sequences and illustrate applications to hyperparameter selection, convergence rate assessment, and quantifying bias-variance tradeoffs in posterior inference. |
---|
Available Formats
Format | Quality | Bitrate | Size | |||
---|---|---|---|---|---|---|
MPEG-4 Video | 640x360 | 1.94 Mbits/sec | 566.09 MB | View | Download | |
WebM * | 640x360 | 548.36 kbits/sec | 156.23 MB | View | Download | |
iPod Video | 480x270 | 521.72 kbits/sec | 148.58 MB | View | Download | |
MP3 | 44100 Hz | 249.74 kbits/sec | 71.22 MB | Listen | Download | |
Auto | (Allows browser to choose a format it supports) |