SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 1912.03500 · 2019

Optimizing Rank-based Metrics with Blackbox Differentiation

arXiv · PDF · Open in the Atlas

Code that ran

We have not lifted any functions out of this paper's repositories yet, so there is nothing we have run. The repositories linked to it are listed below.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Rank-based metrics are some of the most widely used criteria for performance evaluation of computer vision models. Despite years of effort, direct optimization for these metrics remains a challenge due to their non-differentiable and non-decomposable nature. We present an efficient, theoretically sound, and general method for differentiating rank-based metrics with mini-batch gradient descent. In addition, we address optimization instability and sparsity of the supervision signal that both arise from using rank-based metrics as optimization targets. Resulting losses based on recall and Average Precision are applied to image retrieval and object detection tasks. We obtain performance that is competitive with state-of-the-art on standard image retrieval datasets and consistently improve performance of near state-of-the-art object detectors. The code is available at https://github.com/martius-lab/blackbox-backprop

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("1912.03500")
get_code_for_paper("1912.03500")
have("1912.03500")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/1912.03500.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/1912.03500.svg)](https://syntology.ai/paper/1912.03500)

Connect an agent — have() is free.