Attack RMSE Leaderboard: An Introduction and Case Study

TL;DR

This paper introduces techniques for improving leaderboard rankings based on RMSE without using training data, providing insights into model tuning and evaluation strategies.

Contribution

It presents novel tricks for boosting RMSE leaderboard performance without data exploitation, offering practical guidance for model evaluation.

Findings

Effective tricks can improve RMSE scores on leaderboards

Strategies do not rely on training data exploitation

Provides a case study demonstrating these techniques

Abstract

In this manuscript, we briefly introduce several tricks to climb the leaderboards which use RMSE for evaluation without exploiting any training data.