Research map / Time series / Causal and generative time series
Efficient learning of nonlinear prediction models with time-series privileged information
Bastian Jung, Fredrik Daniel Johansson
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 generalized LuPTS with random ReLU features instead of linear LuPTS | generalized LuPTS · Traffic | R2 | Improved |
| add privileged time-series information (generalized LuPTS) to linear learner | linear OLS · Square-Sign | R2 | Hurt |
| add privileged information to random feature learners | random Fourier/ReLU features · Square-Sign | R2 | Improved |
| use nonlinear privileged learners instead of classical counterparts | generalized LuPTS · ADNI and PM air quality | R2 | No clear effect |
| add privileged time-series information to nonlinear learners | generalized LuPTS · Traffic, Square-Sign (averaged over seeds) | sample efficiency R2 | Improved |
| use SRL (stepwise transition objective only) instead of classical representation learner | representation learner · Traffic, Square-Sign, Clocks-LGS | R2 | Hurt |
| add outcome supervision at every time step (CRL/GRL) to SRL (ablation) | SRL · Traffic, Square-Sign, Clocks-LGS | R2 | Improved |
| use generalized distillation for privileged time series instead of classical learning | classical learner · Traffic, Square-Sign, Clocks-LGS | R2 | No clear effect |
| CRL with lambda<1 instead of GRL (lambda=1) (ablation) | CRL · most tasks | R2 | No clear effect |
| add privileged information to classical learner | classical learner · Square-Sign | prediction variance | Improved |
| add privileged information to classical learner | classical learner · Square-Sign | prediction bias | Hurt |
| add privileged supervision (GRL/CRL) to representation learner | classical representation learner · Clocks-LGS | SVCCA latent recovery | Improved |
Closest papers by meaning
- Deep Learning Under Privileged Information Using Heteroscedastic Dropout (CVPR 2018 · similarity 0.85)
- Reduced-Rank Linear Dynamical Systems (AAAI 2018 · similarity 0.85)
- eXponential FAmily Dynamical Systems (XFADS): Large-scale nonlinear Gaussian state-space modeling (NeurIPS 2024 · similarity 0.84)
- LETS Forecast: Learning Embedology for Time Series Forecasting (ICML 2025 · similarity 0.83)
- Probabilistic Decomposed Linear Dynamical Systems for Robust Discovery of Latent Neural Dynamics (NeurIPS 2024 · similarity 0.83)
- Learning Low-dimensional Latent Dynamics from High-dimensional Observations: Non-asymptotics and Lower Bounds (ICML 2024 · similarity 0.83)
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.
