Research map / Probabilistic and scientific modeling / Dynamical systems
Embedding Hybrid Systems into Continuous Latent Vector Fields
Sangli Teng, Hang Liu, Koushil Sreenath
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 |
|---|---|---|---|
| add latent consistency loss L_z (ablation) | CHyLL++ reconstruction loss · Bouncing Ball | reconstruction MSE: 0.738 -> 0.106 | Improved |
| add latent consistency loss L_z (ablation) | CHyLL++ reconstruction loss · 3D Bouncing Ball | reconstruction MSE: 0.416 -> 0.175 | Improved |
| add latent consistency loss L_z (ablation) | CHyLL++ reconstruction loss · Torus | reconstruction MSE: 0.0742 -> 0.0163 | Improved |
| add velocity loss L_v (ablation) | CHyLL++ reconstruction loss · Three-Link Walker | reconstruction MSE: 0.0913 -> 0.118 | Hurt |
| add velocity loss L_v (ablation) | CHyLL++ reconstruction loss · 3D Bouncing Ball | reconstruction MSE: 0.160 -> 0.254 | Hurt |
| add gluing and continuity losses L_g and L_c together (ablation) | CHyLL++ reconstruction loss · Klein Bottle | reconstruction MSE: 0.00954 -> 0.00464 | Improved |
| add gluing and continuity losses L_g and L_c together (ablation) | CHyLL++ reconstruction loss · Three-Link Walker | reconstruction MSE: 0.0731 -> 0.0913 | Hurt |
| use sin activation in last layer instead of ReLU (ablation) | CHyLL++ with L_x,z,c · Torus | reconstruction MSE: 0.00363 -> 0.00304 | Improved |
| use sin activation in last layer instead of ReLU (ablation) | CHyLL++ with L_x,z · Torus | reconstruction MSE: 0.0411 -> 0.0163 | Improved |
| use sin activation in last layer instead of ReLU (ablation) | CHyLL++ with L_x,z · 3D Bouncing Ball | reconstruction MSE: 0.157 -> 0.175 | Hurt |
| raise latent dimension from n+k to 2n (ablation) | CHyLL++ · 3D Bouncing Ball | reconstruction MSE: 0.481 -> 0.160 | Improved |
| raise latent dimension from n+k to 2n (ablation) | CHyLL++ · Klein Bottle | reconstruction MSE: 0.0194 -> 0.00464 | Improved |
Closest papers by meaning
- Anamnesic Neural Differential Equations with Orthogonal Polynomial Projections (ICLR 2023 · similarity 0.86)
- HyRNN: Hybrid Recurrent Neural Networks for Approximating Hybrid Dynamical Systems (AAAI 2026 · similarity 0.86)
- Faithful Embeddings of Irregular and Asynchronous Data for Online Log-NCDEs (NeurIPS 2026 · similarity 0.84)
- Vectorized Conditional Neural Fields: A Framework for Solving Time-dependent Parametric Partial Differential Equations (ICML 2024 · similarity 0.84)
- Identifying latent state transitions in non-linear dynamical systems (ICLR 2025 · similarity 0.83)
- Accurate Differential Operators for Hybrid Neural Fields (CVPR 2025 · 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.
