Research map / Learning theory and algorithms / Prediction and decision losses
TensorFuzz: Debugging Neural Networks with Coverage-Guided Fuzzing
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| guide mutations with activation coverage rather than use random search | MNIST classifier numerical-error search · MNIST classifier with unsafe numerical operations | runs finding a non-finite element: TensorFuzz found a non-finite element every run; Random search was not able to find a non-finite element in any of these | Helped |
| guide mutations with activation coverage rather than use random search | 32-bit and 16-bit MNIST classifiers · MNIST test images within a radius 0.4 infinity-norm ball | examples with discovered prediction disagreement: 70%; Random search was not able to find any disagreements that didn't already exist | Helped |
| fuzz inputs near seed images | 32-bit and 16-bit MNIST classifiers · MNIST test images within a radius 0.4 infinity-norm ball | examples with discovered prediction disagreement: 70% | Helped |
| guide mutations with activation coverage rather than use random search | Tiny Shakespeare character-level language model · Tiny Shakespeare | blacklist words generated: six out of ten words; one | Helped |
| use coverage-guided fuzzing instead of random search | Tiny Shakespeare character-level language model · Tiny Shakespeare | repeat-word behavior found | No clear effect |
Closest papers by meaning
- Cockpit: A Practical Debugging Tool for the Training of Deep Neural Networks (NeurIPS 2021 · similarity 0.83)
- Convolutions and More as Einsum: A Tensor Network Perspective with Advances for Second-Order Methods (NeurIPS 2024 · similarity 0.83)
- Einsum Benchmark: Enabling the Development of Next-Generation Tensor Execution Engines (NeurIPS 2024 · similarity 0.83)
- No Wrong Turns: The Simple Geometry Of Neural Networks Optimization Paths (ICML 2024 · similarity 0.82)
- Accelerating neural network training: An analysis of the AlgoPerf competition (ICLR 2025 · similarity 0.82)
- Tensor Decomposition Meets Knowledge Compilation: A Study Comparing Tensor Trains with OBDDs (AAAI 2025 · similarity 0.82)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.