Research map / Probabilistic and scientific modeling / Molecules and proteins
Meta Learning with Relational Information for Short Sequences
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 HARMLESS (FOMAML) instead of the strongest baseline | HARMLESS · 911-Calls | Log-likelihood: 4.5111±0.3192 -> 4.6362± 0.3241 | Helped |
| Use HARMLESS (MAML) instead of the strongest baseline | HARMLESS · LinkedIn | Log-likelihood: 0.9270 ± 0.0027 -> 1.4070± 0.0105 | Helped |
| Use HARMLESS (Reptile) instead of the strongest baseline | HARMLESS · MathOverflow | Log-likelihood: 1.7805±0.0345 -> 1.8663±0.0342 | Helped |
| Use HARMLESS (Reptile) instead of the strongest baseline | HARMLESS · StackOverflow | Log-likelihood: 1.5594±0.0134 -> 1.6017±0.0097 | Helped |
| Add subject-specific adaptation within each identity (ablation) | HARMLESS (MAML), LinkedIn · LinkedIn | Log-likelihood: 0.9405 ± 0.0032 -> 1.4070± 0.0105; 0.9392 ± 0.0032 -> 1.4070± 0.0105 | Helped |
| Use multiple communities instead of one (ablation) | HARMLESS (MAML), LinkedIn · LinkedIn | Log-likelihood: 0.9432 -> 1.4070 | Helped |
| Use multiple communities instead of one (ablation) | HARMLESS (FOMAML), LinkedIn · LinkedIn | Log-likelihood: 0.9376 -> 1.0129 | Helped |
| Use multiple communities instead of one (ablation) | HARMLESS (Reptile), LinkedIn · LinkedIn | Log-likelihood: 0.9455 -> 0.9540 | Helped |
| Add graph information (ablation) | HARMLESS (MAML), LinkedIn · LinkedIn | Log-likelihood: 0.9507 -> 1.4070 | Helped |
| Add graph information (ablation) | HARMLESS (FOMAML), LinkedIn · LinkedIn | Log-likelihood: 0.9446 -> 1.0129 | Helped |
| Add graph information (ablation) | HARMLESS (Reptile), LinkedIn · LinkedIn | Log-likelihood: 0.9489 -> 0.9540 | Helped |
| Jointly train graph identities and event sequences rather than fixing identities from a separate graph model | HARMLESS (MAML) · Synthetic data | Log-likelihood | Helped |
Closest papers by meaning
- A Dirichlet Mixture Model of Hawkes Processes for Event Sequence Clustering (NeurIPS 2017 · similarity 0.86)
- Learning Hawkes Processes from Short Doubly-Censored Event Sequences (ICML 2017 · similarity 0.86)
- A Hierarchical Bayesian Model for Few-Shot Meta Learning (ICLR 2024 · similarity 0.86)
- Variational Continual Bayesian Meta-Learning (NeurIPS 2021 · similarity 0.85)
- Nonparametric Regressive Point Processes Based on Conditional Gaussian Processes (NeurIPS 2019 · similarity 0.85)
- Transformer Neural Processes: Uncertainty-Aware Meta Learning Via Sequence Modeling (ICML 2022 · similarity 0.85)
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.