Coreset-based Conformal Prediction for large-scale learning

Nery Riquelme-Granada; Khuong An Nguyen; Zhiyuan Luo

Coreset-based Conformal Prediction for large-scale learning

Nery Riquelme-Granada
, Khuong An Nguyen
, Zhiyuan Luo

Research output: Contribution to conference › Paper › peer-review

Abstract

As the volume of data increase rapidly, most traditional machine learning algorithms become computationally prohibitive. Furthermore, the available data can be so big that a single machine's memory can easily be overflown.

We propose Coreset-Based Conformal Prediction, a strategy for dealing with big data by applying conformal predictors to a weighted summary of data - namely the coreset. We compare our approach against stand-alone inductive conformal predictors over three large competition-grade datasets to demonstrate that our coreset-based strategy may not only significantly improve the learning speed, but also retains predictions validity and the predictors' efficiency.

Original language	English
Pages	142-162
Publication status	Published - 2019
Event	8th Symposium on Conformal and Probabilistic Prediction with Applications (COPA 2019) - , Bulgaria Duration: 9 Sept 2019 → 11 Sept 2019 https://cml.rhul.ac.uk/copa2019/

Conference

Conference	8th Symposium on Conformal and Probabilistic Prediction with Applications (COPA 2019)
Country/Territory	Bulgaria
Period	9/09/19 → 11/09/19
Internet address	https://cml.rhul.ac.uk/copa2019/

Keywords

logistic regression
conformal predictors
importance sampling

Access to Document

Cite this

@conference{15f20d8d4c1a4f74abb441aba2724f41,

title = "Coreset-based Conformal Prediction for large-scale learning",

abstract = "As the volume of data increase rapidly, most traditional machine learning algorithms become computationally prohibitive. Furthermore, the available data can be so big that a single machine's memory can easily be overflown. We propose Coreset-Based Conformal Prediction, a strategy for dealing with big data by applying conformal predictors to a weighted summary of data - namely the coreset. We compare our approach against stand-alone inductive conformal predictors over three large competition-grade datasets to demonstrate that our coreset-based strategy may not only significantly improve the learning speed, but also retains predictions validity and the predictors' efficiency.",

keywords = "logistic regression, conformal predictors, importance sampling",

author = "Nery Riquelme-Granada and Nguyen, \{Khuong An\} and Zhiyuan Luo",

year = "2019",

language = "English",

pages = "142--162",

note = "8th Symposium on Conformal and Probabilistic Prediction with Applications (COPA 2019) ; Conference date: 09-09-2019 Through 11-09-2019",

url = "https://cml.rhul.ac.uk/copa2019/",

}

TY - CONF

T1 - Coreset-based Conformal Prediction for large-scale learning

AU - Riquelme-Granada, Nery

AU - Nguyen, Khuong An

AU - Luo, Zhiyuan

PY - 2019

Y1 - 2019

N2 - As the volume of data increase rapidly, most traditional machine learning algorithms become computationally prohibitive. Furthermore, the available data can be so big that a single machine's memory can easily be overflown. We propose Coreset-Based Conformal Prediction, a strategy for dealing with big data by applying conformal predictors to a weighted summary of data - namely the coreset. We compare our approach against stand-alone inductive conformal predictors over three large competition-grade datasets to demonstrate that our coreset-based strategy may not only significantly improve the learning speed, but also retains predictions validity and the predictors' efficiency.

AB - As the volume of data increase rapidly, most traditional machine learning algorithms become computationally prohibitive. Furthermore, the available data can be so big that a single machine's memory can easily be overflown. We propose Coreset-Based Conformal Prediction, a strategy for dealing with big data by applying conformal predictors to a weighted summary of data - namely the coreset. We compare our approach against stand-alone inductive conformal predictors over three large competition-grade datasets to demonstrate that our coreset-based strategy may not only significantly improve the learning speed, but also retains predictions validity and the predictors' efficiency.

KW - logistic regression

KW - conformal predictors

KW - importance sampling

M3 - Paper

SP - 142

EP - 162

T2 - 8th Symposium on Conformal and Probabilistic Prediction with Applications (COPA 2019)

Y2 - 9 September 2019 through 11 September 2019

ER -

Coreset-based Conformal Prediction for large-scale learning

Abstract

Conference

Keywords

Access to Document

Fingerprint

Cite this