Research map / Learning theory and algorithms / Submodular and game algorithms
Scalable Deletion-Robust Submodular Maximization: Data Summarization with Privacy and Fairness Constraints
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 |
|---|---|---|---|
| Use ROBUST-CENTRALIZED/ROBUST-STREAMING to select features after deleting sensitive features | feature selection with fairness constraints · Adult Income, SVM | SVM accuracy: drop of 0.2% vs GREEDYD (0.793 to 0.791) | No clear effect |
| Use ROBUST-CENTRALIZED/ROBUST-STREAMING for feature selection after deleting sensitive features | feature selection with fairness constraints · Adult Income, Naive Bayes | Naive Bayes accuracy: 0.781 vs GREEDYD 0.781 | No clear effect |
| Use ROBUST-CENTRALIZED instead of ROBUST for feature selection under deletions | feature selection · Adult Income, memory | number of stored items: 22 vs 39 stored items | Helped |
| Use ROBUST-CENTRALIZED/ROBUST-STREAMING instead of ROBUST and SG under increasing deletions | Uber location summarization · Uber pickups, k=20, d=5 | normalized objective value | Helped |
| Use ROBUST-CENTRALIZED instead of ROBUST for memory footprint | Uber location summarization · Uber pickups memory complexity | items kept in memory | Helped |
| Use ROBUST-DISTRIBUTED instead of distributed stochastic greedy for summarization under deletions | Census1990 large-scale summarization · Census1990, D1-D4 deletion strategies, k=50 and k=100 | normalized objective value | Helped |
| Use ROBUST-DISTRIBUTED with 12 machines on Census1990 | Census1990 large-scale summarization · 80% random deletion (D2) | normalized objective value | Helped |
| Increase number of deletions beyond designed d for ROBUST algorithms | Uber and Adult Income experiments · deletions r much greater than d | normalized objective value | No clear effect |
Closest papers by meaning
- Deletion-Robust Submodular Maximization: Data Summarization with “the Right to be Forgotten” (ICML 2017 · similarity 0.94)
- Differentially Private Submodular Maximization: Data Summarization in Disguise (ICML 2017 · similarity 0.94)
- Streaming Robust Submodular Maximization: A Partitioned Thresholding Approach (NeurIPS 2017 · similarity 0.89)
- Fairness in Streaming Submodular Maximization: Algorithms and Hardness (NeurIPS 2020 · similarity 0.89)
- Deletion-Robust Submodular Maximization with Knapsack Constraints (AAAI 2024 · similarity 0.88)
- Deletion Robust Submodular Maximization over Matroids (ICML 2022 · similarity 0.88)
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.