# Semantic-Based Neural Network Repair

arXiv: https://arxiv.org/abs/2306.07995 · HTML: https://mlsanity.com/paper/2306.07995/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- Builds on 19 of its 62 references that are in the map.
- First posted on arXiv 12 Jun 2023.

## Builds on (most cited first, 19 of 19)

- [Fashion-MNIST: a Novel Image Dataset for Benchmarking Machine Learning Algorithms](https://mlsanity.com/paper/1708.07747/) (2017)
- [Deep Learning in Neural Networks: An Overview](https://mlsanity.com/paper/1404.7828/) (2014)
- [AutoML: A Survey of the State-of-the-Art](https://mlsanity.com/paper/1908.00709/) (2019)
- [Visual Analytics in Deep Learning: An Interrogative Survey for the Next Frontiers](https://mlsanity.com/paper/1801.06889/) (2018)
- [Taxonomy of Real Faults in Deep Learning Systems](https://mlsanity.com/paper/1910.11015/) (2019)
- [Seq2Seq-Vis: A Visual Debugging Tool for Sequence-to-Sequence Models](https://mlsanity.com/paper/1804.09299/) (2018)
- [TensorFuzz: Debugging Neural Networks with Coverage-Guided Fuzzing](https://mlsanity.com/paper/1807.10875/) (2018)
- [Automated Machine Learning: From Principles to Practices](https://mlsanity.com/paper/1810.13306/) (2018)
- [Causality-based Neural Network Repair](https://mlsanity.com/paper/2204.09274/) (2022)
- [Provable Repair of Deep Neural Networks](https://mlsanity.com/paper/2104.04413/) (2021)
- [Repairing Deep Neural Networks: Fix Patterns and Challenges](https://mlsanity.com/paper/2005.00972/) (2020)
- [Identifying Implementation Bugs in Machine Learning based Image Classifiers using Metamorphic Testing](https://mlsanity.com/paper/1808.05353/) (2018)
- [NNrepair: Constraint-based Repair of Neural Network Classifiers](https://mlsanity.com/paper/2103.12535/) (2021)
- [DocTer: Documentation Guided Fuzzing for Testing Deep Learning API Functions](https://mlsanity.com/paper/2109.01002/) (2021)
- [Neural Network Repair with Reachability Analysis](https://mlsanity.com/paper/2108.04214/) (2021)
- [DeepDiagnosis: Automatically Diagnosing Faults and Recommending Actionable Fixes in Deep Learning Programs](https://mlsanity.com/paper/2112.04036/) (2021)
- [SOCRATES: Towards a Unified Platform for Neural Network Analysis](https://mlsanity.com/paper/2007.11206/) (2020)
- [A new perspective on the competent programmer hypothesis through the reproduction of bugs with repeated mutations](https://mlsanity.com/paper/2104.02517/) (2021)
- [ExAIS: Executable AI Semantics](https://mlsanity.com/paper/2202.09868/) (2022)

## Built on by (most cited first, 3 of 3)

- [LUNA: A Model-Based Universal Analysis Framework for Large Language Models](https://mlsanity.com/paper/2310.14211/) (2023)
- [WARP: Guaranteed Inner-Layer Repair of NLP Transformers](https://mlsanity.com/paper/2604.00938/) (2026)
- [DeepProv: Behavioral Characterization and Repair of Neural Networks via Inference Provenance Graph Analysis](https://mlsanity.com/paper/2509.26562/) (2025)

## Abstract (arXiv)

Recently, neural networks have spread into numerous fields including many safety-critical systems. Neural networks are built (and trained) by programming in frameworks such as TensorFlow and PyTorch. Developers apply a rich set of pre-defined layers to manually program neural networks or to automatically generate them (e.g., through AutoML). Composing neural networks with different layers is error-prone due to the non-trivial constraints that must be satisfied in order to use those layers. In this work, we propose an approach to automatically repair erroneous neural networks. The challenge is in identifying a minimal modification to the network so that it becomes valid. Modifying a layer might have cascading effects on subsequent layers and thus our approach must search recursively to identify a "globally" minimal modification. Our approach is based on an executable semantics of deep learning layers and focuses on four kinds of errors which are common in practice. We evaluate our approach for two usage scenarios, i.e., repairing automatically generated neural networks and manually written ones suffering from common model bugs. The results show that we are able to repair 100% of a set of randomly generated neural networks (which are produced with an existing AI framework testing approach) effectively and efficiently (with an average repair time of 21.08s) and 93.75% of a collection of real neural network bugs (with an average time of 3min 40s).