Cost-Effective HITs for Relative Similarity Comparisons

TL;DR

This paper investigates cost-effective methods for collecting triplet similarity comparisons in machine learning, emphasizing UI design improvements over sampling algorithms to enhance data quality and reduce costs.

Contribution

It introduces UI design strategies for triplet collection tasks that improve data quality and reduce costs, alongside providing a new dataset and labels from crowd workers.

Findings

UI improvements significantly increase data quality

Cost reduction achieved through better task display

Provided a new dataset with crowd-sourced triplet labels

Abstract

Similarity comparisons of the form "Is object a more similar to b than to c?" are useful for computer vision and machine learning applications. Unfortunately, an embedding of $n$ points is specified by $n^3$ triplets, making collecting every triplet an expensive task. In noticing this difficulty, other researchers have investigated more intelligent triplet sampling techniques, but they do not study their effectiveness or their potential drawbacks. Although it is important to reduce the number of collected triplets, it is also important to understand how best to display a triplet collection task to a user. In this work we explore an alternative display for collecting triplets and analyze the monetary cost and speed of the display. We propose best practices for creating cost effective human intelligence tasks for collecting triplets. We show that rather than changing the sampling algorithm, simple changes to the crowdsourcing UI can lead to much higher quality embeddings. We also provide a dataset as well as the labels collected from crowd workers.