Research map / Learning theory and algorithms / Sample complexity
f-INE: A Hypothesis Testing Framework for Estimating Influence under Training Randomness
Subhodip Panda, Dhruv Tarsadiya, Shashwat Sourav, Prathosh AP, Sai Praneeth Karimireddy
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| use hypothesis-testing f-INE influence instead of mean-based LESS | f-INE · LIMA poisoning, Joe Biden, top 20% ranking | recall of poisoned instructions (%): 44% -> more than 60% | Improved |
| compare full distributions (f-INE) instead of means (LESS) | f-INE · LIMA poisoning variability across 3 runs | average coefficient of variation: 64% lower than LESS at p=1.0 | Improved |
| f-INE vs TraceIn for mislabel detection on MNIST | f-INE · MNIST mislabeled detection | recall: +0.05% vs TraceIn | No clear effect |
| f-INE vs TRAK and Influence Functions on MNIST mislabel detection | f-INE · MNIST mislabeled detection | recall: +13.85% over TRAK, +3.83% over IFs | Improved |
| decrease gradient projection dimension (from 8192 down to 1024) (ablation) | f-INE · LIMA poisoning | utility (recall of poisoned instructions) | Hurt |
| use more checkpoints for gradient collection (60, 80 vs 100) (ablation) | f-INE · LIMA poisoning | utility | Improved |
Closest papers by meaning
- The Mirrored Influence Hypothesis: Efficient Data Influence Estimation by Harnessing Forward Passes (CVPR 2024 · similarity 0.89)
- DataInf: Efficiently Estimating Data Influence in LoRA-tuned LLMs and Diffusion Models (ICLR 2024 · similarity 0.87)
- Towards Robust Influence Functions with Flat Validation Minima (ICML 2025 · similarity 0.87)
- Neural Networks for Learnable and Scalable Influence Estimation of Instruction Fine-Tuning Data (NeurIPS 2025 · similarity 0.87)
- A Versatile Influence Function for Data Attribution with Non-Decomposable Loss (ICML 2025 · similarity 0.87)
- Rescaled Influence Functions: Accurate Data Attribution in High Dimension (NeurIPS 2025 · similarity 0.87)
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.
