Trojan Horse Hunt in Time Series Forecasting for Space Operations

2025-06-02

AI Summary

Satellite telemetry forecasting models for safety-critical space missions are vulnerable to adversarial data poisoning—specifically, backdoor attacks—yet no effective detection or reconstruction method exists for time-series prediction models. Method: We propose the first backdoor trigger detection and reconstruction framework tailored to time-series models, addressing 45 unknown multivariate triggers of arbitrary shape, amplitude, and duration. Our approach integrates N-HiTS-based poisoned model analysis, gradient-guided temporal localization, multi-scale inversion optimization, and telemetry-informed prior-constrained reconstruction—overcoming structural limitations of image-domain methods (e.g., Neural Cleanse) in modeling temporal dependencies. Contribution/Results: This work pioneers systematic backdoor detection in time-series domains; it precisely reconstructs all 45 triggers and achieves multiple Top 10% rankings in Kaggle competitions. The framework significantly enhances the trustworthiness and robustness of AI models deployed in aerospace applications.

Technology Category

• Machine Learning: Time-Series/Data Streams
• Computer Vision: Adversarial Attacks & Robustness
• Planning, Routing, and Scheduling: Model-Based Reasoning

Application Category

• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

This competition hosted on Kaggle (https://www.kaggle.com/competitions/trojan-horse-hunt-in-space) is the first part of a series of follow-up competitions and hackathons related to the"Assurance for Space Domain AI Applications"project funded by the European Space Agency (https://assurance-ai.space-codev.org/). The competition idea is based on one of the real-life AI security threats identified within the project -- the adversarial poisoning of continuously fine-tuned satellite telemetry forecasting models. The task is to develop methods for finding and reconstructing triggers (trojans) in advanced models for satellite telemetry forecasting used in safety-critical space operations. Participants are provided with 1) a large public dataset of real-life multivariate satellite telemetry (without triggers), 2) a reference model trained on the clean data, 3) a set of poisoned neural hierarchical interpolation (N-HiTS) models for time series forecasting trained on the dataset with injected triggers, and 4) Jupyter notebook with the training pipeline and baseline algorithm (the latter will be published in the last month of the competition). The main task of the competition is to reconstruct a set of 45 triggers (i.e., short multivariate time series segments) injected into the training data of the corresponding set of 45 poisoned models. The exact characteristics (i.e., shape, amplitude, and duration) of these triggers must be identified by participants. The popular Neural Cleanse method is adopted as a baseline, but it is not designed for time series analysis and new approaches are necessary for the task. The impact of the competition is not limited to the space domain, but also to many other safety-critical applications of advanced time series analysis where model poisoning may lead to serious consequences.

Problem

• Detect adversarial triggers in satellite telemetry forecasting models
• Reconstruct multivariate time series triggers in poisoned models
• Improve Neural Cleanse for time series security applications

Innovation

• Detect triggers in satellite telemetry forecasting models
• Reconstruct adversarial triggers in time series data
• Improve Neural Cleanse for time series analysis