Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Bayesian Learning from Sequential Data using Gaussian Processes with Signature Covariances
Csaba Toth, Harald Oberhauser
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 signature covariance instead of the alternative GP covariance baselines | GP classifier · TS classification, all datasets | mean rank (nlpp): 4.75 -> 2.44 | Improved |
| Use signature covariance instead of the alternative GP covariance baselines | GP classifier · TS classification, all datasets | mean rank (accuracy): 4.22 -> 2.88 | Improved |
| Add an LSTM sequence transformation before the signature GP (ablation) | GP-Sig · TS classification, n_X < 300 | mean rank (accuracy): 2.80 -> 3.00 | Hurt |
| Add an LSTM sequence transformation before the signature GP (ablation) | GP-Sig · TS classification, n_X ≥ 300 | mean rank (accuracy): 3.00 -> 2.17 | Improved |
| Use an LSTM rather than a GRU sequence transformation with the signature GP (ablation) | GP-Sig-RNN · TS classification, all datasets | mean rank (accuracy): 3.25 -> 2.69 | Improved |
| Replace inducing tensors with inducing sequences at matched computational complexity (ablation) | GP-Sig · AUSLAN | accuracy | Hurt |
| Lift observations through a nonlinear feature map before computing signatures (ablation) | Signature covariance · Multivariate time-series classification | accuracy | Improved |
| Add lags before computing signatures (ablation) | GP-Sig · Multivariate time-series classification | accuracy | Improved |
Closest papers by meaning
- SigGPDE: Scaling Sparse Gaussian Processes on Sequential Data (ICML 2021 · similarity 0.89)
- Memory-Based Dual Gaussian Processes for Sequential Learning (ICML 2023 · similarity 0.89)
- Beyond the Mean-Field: Structured Deep Gaussian Processes Improve the Predictive Uncertainties (NeurIPS 2020 · similarity 0.88)
- Learning with Expected Signatures: Theory and Applications (ICML 2025 · similarity 0.87)
- Learning Compositional Sparse Gaussian Processes with a Shrinkage Prior (AAAI 2021 · similarity 0.87)
- Sparse Variational Student-t Processes (AAAI 2024 · similarity 0.86)
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.
