Solving the Families In the Wild Kinship Verification Challenge by Program Synthesis

TL;DR

This paper presents a high-performing kinship verification method that leverages ensemble models including human expertise and OpenAI Codex, achieving top results in a major challenge.

Contribution

It introduces the novel use of foundation models like OpenAI Codex for generating kinship verification models and programs, enhancing accuracy in the field.

Findings

Achieved top 3 placement in the 2021 Recognizing Families in the Wild challenge.

Demonstrated Codex's ability to generate effective kinship verification models.

Ensembled human and foundation model outputs for improved performance.

Abstract

Kinship verification is the task of determining whether a parent-child, sibling, or grandparent-grandchild relationship exists between two people and is important in social media applications, forensic investigations, finding missing children, and reuniting families. We demonstrate high quality kinship verification by participating in the 2021 Recognizing Families in the Wild challenge which provides the largest publicly available dataset in the field. Our approach is among the top 3 winning entries in the competition. We ensemble models written by both human experts and a foundation model, OpenAI Codex, trained on text and code. We use Codex to generate model variants, and also demonstrate its ability to generate entire running programs for kinship verification tasks of specific relationships.