Research map / Label-efficient and robust learning / Recognition and long-tail
Identifying Unknown Unknowns in the Open World: Representations and Policies for Guided Exploration
Himabindu Lakkaraju, Ece Kamar, Rich Caruana, Eric Horvitz
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 DSP partitions instead of k-means partitions based on features, confidence scores, or both | Unknown-unknown discovery · Sentiment Snippets, Subjectivity, Amazon Reviews, Image Data | entropy of unknown-unknown distribution across partitions | Improved |
| Jointly partition using features and confidence scores rather than only features or only confidence scores (ablation) | DSP · Sentiment Snippets, Subjectivity, Amazon Reviews, Image Data | entropy of unknown-unknown distribution across partitions | Improved |
| Use DSP instead of random partitioning | Unknown-unknown discovery · Sentiment Snippets, Subjectivity, Amazon Reviews, Image Data | entropy of unknown-unknown distribution across partitions: 32-37% lower | Improved |
| Use UUB instead of random, greedy, epsilon-greedy, UCB, UCB_f, sliding-window, or discounted UCB policies | DSP partitions · Image Data | cumulative regret | Improved |
| Use UUB instead of the compared bandit policies | DSP partitions · Sentiment Snippets and Subjectivity | cumulative regret | Improved |
| Use UUB instead of discounted UCB with discounting factor 0.5 | DSP partitions · Amazon Reviews | cumulative regret | Hurt |
| Use the complete DSP + UUB pipeline instead of random sampling, least-average-similarity, least-maximum-similarity, or most-uncertain sampling | Unknown-unknown discovery · Image Data | cumulative regret | Improved |
Closest papers by meaning
- Active Target Discovery under Uninformative Priors: The Power of Permanent and Transient Memory (NeurIPS 2025 · similarity 0.83)
- Can In-Context Learning Support Intrinsic Curiosity? (NeurIPS 2026 · similarity 0.83)
- FIND: A Framework for Discovering Formulas in Data (AAAI 2025 · similarity 0.82)
- Looking Beyond the Known: Towards a Data Discovery Guided Open-World Object Detection (NeurIPS 2025 · similarity 0.82)
- Rethinking Epistemic and Aleatoric Uncertainty for Active Open-Set Annotation: An Energy-Based Approach (CVPR 2025 · similarity 0.82)
- Learning from Uncertain Data: From Possible Worlds to Possible Models (NeurIPS 2024 · 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.
