Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Memory-Based Dual Gaussian Processes for Sequential Learning
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 |
|---|---|---|---|
| Add a representative memory of past examples to Dual-SVGP (ablation) | Dual-SVGP · Split MNIST | accuracy: $.208 (.007)$ -> $.909 (.001)$ | Helped |
| Add a representative memory of past examples to Dual-SVGP (ablation) | Dual-SVGP · Split MNIST | NLPD: 68.038 (2.830) -> $\phantom{0}0.316 (0.010)$ | Helped |
| Use BLS-weighted selection rather than random selection for examples moved from training to test (ablation) | Memory selection on Split MNIST · Split MNIST | accuracy | Hurt |
| Increase memory size using BLS-selected examples (ablation) | Dual-SVGP · Split MNIST | accuracy | Helped |
| Use memory-based sequential Dual-SVGP rather than OVC | Streaming GP · UCI data sets | NLPD: .57 (.02) vs .57 (.02); .44 (.03) vs .49 (.03); .41 (.05) vs .42 (.05); .26 (.03) vs .28 (.04); .02 (.00) vs .08 (.01); | Helped |
| Use memory-based posterior updating rather than OVC | SVGP · Schistosomiasis hotspot modelling | MSE | Helped |
| Use memory-based posterior updating rather than OVC | SVGP · Schistosomiasis hotspot modelling | accuracy | Helped |
| Use memory-based posterior updating rather than OVC | SVGP · Schistosomiasis hotspot modelling | time per step: 85 seconds (standard deviation 11 seconds) -> 62 seconds (standard deviation 7 seconds) | Helped |
| Use memory-based Dual-SVGP rather than OVC for batch Bayesian optimization | Dual-SVGP · Lunar lander | reward | Helped |
| Use memory-based Dual-SVGP rather than the cited method | Dual-SVGP · Sequential magnetic anomaly mapping | representation quality | Helped |
| Use memory-based Dual-SVGP rather than the cited continual-learning baseline | Dual-SVGP · Split MNIST | accuracy: $.909 (.001)$ vs $.905 (.010)$ | Helped |
| Use natural-gradient optimization for variational parameters rather than the compared inference method | Dual-SVGP · Adult | inference wall-clock time | Helped |
Closest papers by meaning
- Variational Auto-Regressive Gaussian Processes for Continual Learning (ICML 2021 · similarity 0.90)
- Orthogonally Decoupled Variational Gaussian Processes (NeurIPS 2018 · similarity 0.89)
- New Bounds for Sparse Variational Gaussian Processes (ICML 2025 · similarity 0.89)
- Recurrent Memory for Online Interdomain Gaussian Processes (NeurIPS 2025 · similarity 0.88)
- Sparse Gaussian Processes: Structured Approximations and Power-EP Revisited (NeurIPS 2025 · similarity 0.87)
- Sparse Variational Student-t Processes (AAAI 2024 · 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.