Trojan horse hunt in deep forecasting models: Insights from the European Space Agency competition

TL;DR

This paper discusses the security risks of trojan horse attacks in deep forecasting models used in space applications, presenting a competition-based approach to identify hidden triggers in models.

Contribution

It introduces a novel task formulation, benchmark, and evaluation protocol for detecting trojan triggers in deep forecasting models, along with insights from a large-scale competition.

Findings

Effective trigger identification methods were developed.

Key insights into trojan attack characteristics in time series models.

Benchmark dataset and evaluation protocol established.

Abstract

Forecasting plays a crucial role in modern safety-critical applications, such as space operations. However, the increasing use of deep forecasting models introduces a new security risk of trojan horse attacks, carried out by hiding a backdoor in the training data or directly in the model weights. Once implanted, the backdoor is activated by a specific trigger pattern at test time, causing the model to produce manipulated predictions. We focus on this issue in our \textit{Trojan Horse Hunt} data science competition, where more than 200 teams faced the task of identifying triggers hidden in deep forecasting models for spacecraft telemetry. We describe the novel task formulation, benchmark set, evaluation protocol, and best solutions from the competition. We further summarize key insights and research directions for effective identification of triggers in time series forecasting models. All materials are publicly available on the official competition webpage https://www.kaggle.com/competitions/trojan-horse-hunt-in-space.