# Model Assertions for Monitoring and Improving ML Models

arXiv: https://arxiv.org/abs/2003.01668 · HTML: https://mlsanity.com/paper/2003.01668/

## Key facts

- Cited by 22 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 3 Mar 2020.

## Built on by (most cited first, 22 of 22)

- [Collaboration Challenges in Building ML-Enabled Systems: Communication, Documentation, Engineering, and Process](https://mlsanity.com/paper/2110.10234/) (2021)
- [DSPy Assertions: Computational Constraints for Self-Refining Language Model Pipelines](https://mlsanity.com/paper/2312.13382/) (2023)
- [SPADE: Synthesizing Data Quality Assertions for Large Language Model Pipelines](https://mlsanity.com/paper/2401.03038/) (2024)
- [Benchmarking Safety Monitors for Image Classifiers with Machine Learning](https://mlsanity.com/paper/2110.01232/) (2021)
- [Swiss Cheese Model for AI Safety: A Taxonomy and Reference Architecture for Multi-Layered Guardrails of Foundation Model Based Agents](https://mlsanity.com/paper/2408.02205/) (2024)
- [Did the Model Change? Efficiently Assessing Machine Learning API Shifts](https://mlsanity.com/paper/2107.14203/) (2021)
- [Trust in AI: Interpretability is not necessary or sufficient, while black-box interaction is necessary and sufficient](https://mlsanity.com/paper/2202.05302/) (2022)
- [Enabling SQL-based Training Data Debugging for Federated Learning](https://mlsanity.com/paper/2108.11884/) (2021)
- [Reproducibility, Replicability and Beyond: Assessing Production Readiness of Aspect Based Sentiment Analysis in the Wild](https://mlsanity.com/paper/2101.09449/) (2021)
- [HAPI: A Large-scale Longitudinal Dataset of Commercial ML API Predictions](https://mlsanity.com/paper/2209.08443/) (2022)
- [An investigation of challenges encountered when specifying training data and runtime monitors for safety critical ML applications](https://mlsanity.com/paper/2301.13476/) (2023)
- [Framework for Certification of AI-Based Systems](https://mlsanity.com/paper/2302.11049/) (2023)
- [Safety Monitoring of Machine Learning Perception Functions: a Survey](https://mlsanity.com/paper/2412.06869/) (2024)
- [Explainability-aided Domain Generalization for Image Classification](https://mlsanity.com/paper/2104.01742/) (2021)
- [DEBUG-HD: Debugging TinyML models on-device using Hyper-Dimensional computing](https://mlsanity.com/paper/2411.10692/) (2024)
- [Self-Aware Object Detection via Degradation Manifolds](https://mlsanity.com/paper/2602.18394/) (2026)
- [MGit: A Model Versioning and Management System](https://mlsanity.com/paper/2307.07507/) (2023)
- [Semantic Integrity Constraints: Declarative Guardrails for AI-Augmented Data Processing Systems](https://mlsanity.com/paper/2503.00600/) (2025)
- [Graph Lineages and Skeletal Graph Products](https://mlsanity.com/paper/2508.00197/) (2025)
- [Perspectives on a Reliability Monitoring Framework for Agentic AI Systems](https://mlsanity.com/paper/2511.09178/) (2025)
- [The Machine Learning Canvas: Empirical Findings on Why Strategy Matters More Than AI Code Generation](https://mlsanity.com/paper/2601.01839/) (2026)
- [Knowledge-Guided Failure Prediction: Detecting When Object Detectors Miss Safety-Critical Objects](https://mlsanity.com/paper/2603.25499/) (2026)

## Abstract (arXiv)

ML models are increasingly deployed in settings with real world interactions such as vehicles, but unfortunately, these models can fail in systematic ways. To prevent errors, ML engineering teams monitor and continuously improve these models. We propose a new abstraction, model assertions, that adapts the classical use of program assertions as a way to monitor and improve ML models. Model assertions are arbitrary functions over a model's input and output that indicate when errors may be occurring, e.g., a function that triggers if an object rapidly changes its class in a video. We propose methods of using model assertions at all stages of ML system deployment, including runtime monitoring, validating labels, and continuously improving ML models. For runtime monitoring, we show that model assertions can find high confidence errors, where a model returns the wrong output with high confidence, which uncertainty-based monitoring techniques would not detect. For training, we propose two methods of using model assertions. First, we propose a bandit-based active learning algorithm that can sample from data flagged by assertions and show that it can reduce labeling costs by up to 40% over traditional uncertainty-based methods. Second, we propose an API for generating "consistency assertions" (e.g., the class change example) and weak labels for inputs where the consistency assertions fail, and show that these weak labels can improve relative model quality by up to 46%. We evaluate model assertions on four real-world tasks with video, LIDAR, and ECG data.