Saved in:
Bibliographic Details
Main Authors: Dinitz, Michael, Im, Sungjin, Lavastida, Thomas, Moseley, Benjamin, Niaparast, Aidin, Vassilvitskii, Sergei
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2411.16030
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910712811487232
author Dinitz, Michael
Im, Sungjin
Lavastida, Thomas
Moseley, Benjamin
Niaparast, Aidin
Vassilvitskii, Sergei
author_facet Dinitz, Michael
Im, Sungjin
Lavastida, Thomas
Moseley, Benjamin
Niaparast, Aidin
Vassilvitskii, Sergei
contents Algorithms with (machine-learned) predictions is a powerful framework for combining traditional worst-case algorithms with modern machine learning. However, the vast majority of work in this space assumes that the prediction itself is non-probabilistic, even if it is generated by some stochastic process (such as a machine learning system). This is a poor fit for modern ML, particularly modern neural networks, which naturally generate a distribution. We initiate the study of algorithms with distributional predictions, where the prediction itself is a distribution. We focus on one of the simplest yet fundamental settings: binary search (or searching a sorted array). This setting has one of the simplest algorithms with a point prediction, but what happens if the prediction is a distribution? We show that this is a richer setting: there are simple distributions where using the classical prediction-based algorithm with any single prediction does poorly. Motivated by this, as our main result, we give an algorithm with query complexity $O(H(p) + \log η)$, where $H(p)$ is the entropy of the true distribution $p$ and $η$ is the earth mover's distance between $p$ and the predicted distribution $\hat p$. This also yields the first distributionally-robust algorithm for the classical problem of computing an optimal binary search tree given a distribution over target keys. We complement this with a lower bound showing that this query complexity is essentially optimal (up to constants), and experiments validating the practical usefulness of our algorithm.
format Preprint
id arxiv_https___arxiv_org_abs_2411_16030
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Binary Search with Distributional Predictions
Dinitz, Michael
Im, Sungjin
Lavastida, Thomas
Moseley, Benjamin
Niaparast, Aidin
Vassilvitskii, Sergei
Machine Learning
Data Structures and Algorithms
Algorithms with (machine-learned) predictions is a powerful framework for combining traditional worst-case algorithms with modern machine learning. However, the vast majority of work in this space assumes that the prediction itself is non-probabilistic, even if it is generated by some stochastic process (such as a machine learning system). This is a poor fit for modern ML, particularly modern neural networks, which naturally generate a distribution. We initiate the study of algorithms with distributional predictions, where the prediction itself is a distribution. We focus on one of the simplest yet fundamental settings: binary search (or searching a sorted array). This setting has one of the simplest algorithms with a point prediction, but what happens if the prediction is a distribution? We show that this is a richer setting: there are simple distributions where using the classical prediction-based algorithm with any single prediction does poorly. Motivated by this, as our main result, we give an algorithm with query complexity $O(H(p) + \log η)$, where $H(p)$ is the entropy of the true distribution $p$ and $η$ is the earth mover's distance between $p$ and the predicted distribution $\hat p$. This also yields the first distributionally-robust algorithm for the classical problem of computing an optimal binary search tree given a distribution over target keys. We complement this with a lower bound showing that this query complexity is essentially optimal (up to constants), and experiments validating the practical usefulness of our algorithm.
title Binary Search with Distributional Predictions
topic Machine Learning
Data Structures and Algorithms
url https://arxiv.org/abs/2411.16030