# Mechanistic Interpretability for Learning Assurance of a Vision-Based Landing System

arXiv: https://arxiv.org/abs/2605.20607 · HTML: https://mlsanity.com/paper/2605.20607/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Placed in Circuits and mechanistic analysis, one of 13 directions in Mechanistic interpretability.
- Builds on 17 of its 28 references that are in the map.
- First posted on arXiv 20 May 2026.

## Placed in

- Mechanistic interpretability / [Circuits and mechanistic analysis](https://mlsanity.com/mechinterp/circuits-and-mechanistic-analysis/)

## Builds on (most cited first, 17 of 17)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://mlsanity.com/paper/2010.11929/) (2020)
- [DINOv2: Learning Robust Visual Features without Supervision](https://mlsanity.com/paper/2304.07193/) (2023)
- [Emerging Properties in Self-Supervised Vision Transformers](https://mlsanity.com/paper/2104.14294/) (2021)
- [Shortcut Learning in Deep Neural Networks](https://mlsanity.com/paper/2004.07780/) (2020)
- [Sparse Autoencoders Find Highly Interpretable Features in Language Models](https://mlsanity.com/paper/2309.08600/) (2023)
- [Toy Models of Superposition](https://mlsanity.com/paper/2209.10652/) (2022)
- [End-to-End Training of Deep Visuomotor Policies](https://mlsanity.com/paper/1504.00702/) (2015)
- [The Linear Representation Hypothesis and the Geometry of Large Language Models](https://mlsanity.com/paper/2311.03658/) (2023)
- [ViTPose: Simple Vision Transformer Baselines for Human Pose Estimation](https://mlsanity.com/paper/2204.12484/) (2022)
- [Finding Neurons in a Haystack: Case Studies with Sparse Probing](https://mlsanity.com/paper/2305.01610/) (2023)
- [Integral Human Pose Regression](https://mlsanity.com/paper/1711.08229/) (2017)
- [Open Problems in Mechanistic Interpretability](https://mlsanity.com/paper/2501.16496/) (2025)
- [Human Pose Regression by Combining Indirect Part Detection and Contextual Information](https://mlsanity.com/paper/1710.02322/) (2017)
- [LARD 2.0: Enhanced Datasets and Benchmarking for Autonomous Landing Systems](https://mlsanity.com/paper/2603.26748/) (2026)
- [Predictive Uncertainty for Runtime Assurance of a Real-Time Computer Vision-Based Landing System](https://mlsanity.com/paper/2508.09732/) (2025)
- [DB-KSVD: Scalable Alternating Optimization for Disentangling High-Dimensional Embedding Spaces](https://mlsanity.com/paper/2505.18441/) (2025)

## Built on by (most cited first, 1 of 1)

- [Robust Validation to Geometric Perturbations for Autonomous Pose Estimation](https://mlsanity.com/paper/2608.21066/) (2026)

## Abstract (arXiv)

EASA's learning-assurance guidance requires data-driven aviation systems to build and monitor their own situation representation, yet for neural networks the technical means to provide such evidence remain an open problem. We address this gap for a vision-based aircraft landing system: we propose that a minimally assurable model must at least be shown to separate content from style in its own situation representation. Showing that the model's predictions then rely largely on the contentful representation components leads to a concrete assurance path. To demonstrate this assurance path on a concrete model we train a vision transformer model for runway keypoint regression on the LARDv2 dataset. The model, which acts as the subject for our assurance demonstration, produces per-patch embeddings that we decompose into interpretable atoms via K-SVD sparse dictionary learning. A qualitative visualization confirms that contentful atoms track task-relevant runway structure and stylistic atoms track domain-specific appearance, and the regression head is shown to place almost all of its linear weight on contentful atoms. We further build on the content/style separation and define out-of-model-scope (OOMS) detection, a novel runtime assurance approach directly monitoring the model's situation representation. OOMS monitoring is complementary to operational design domain and output-space out-of-distribution monitoring and addresses concrete requirements of the recent EASA guidance. By directly analyzing a model's situation representation both at test time and runtime, this work delivers the first concrete piece of the representation-level evidence that EASA learning-assurance guidance demands, and points to mechanistic interpretability as a practical building block of future aviation safety cases.