Research map / Time series / Causal and generative time series
Liquid Time-constant Networks
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 LTC instead of CT-GRU | Time-series prediction model · Gesture | accuracy: 68.31% $\pm$ 1.78 -> 69.55% $\pm$ 1.13 | Helped |
| Use LTC instead of CT-RNN | Time-series prediction model · Occupancy | accuracy: 94.54% $\pm$ 0.54 -> 94.63% $\pm$ 0.17 | Helped |
| Use LTC instead of Neural ODE | Time-series prediction model · Activity recognition | accuracy: 97.26\% $\pm$ 0.10 -> 95.67% $\pm$ 0.575 | Hurt |
| Use LTC instead of LSTM | Time-series prediction model · Sequential MNIST | accuracy: 98.41\% $\pm$ 0.12 -> 97.57% $\pm$ 0.18 | Hurt |
| Use LTC instead of LSTM | Time-series prediction model · Traffic | squared error: 0.169 $\pm$ 0.004 -> 0.099 $\pm$ 0.0095 | Helped |
| Use LTC instead of CT-GRU | Time-series prediction model · Power | squared-error: 0.586 $\pm$ 0.003 -> 0.642 $\pm$ 0.021 | Hurt |
| Use LTC instead of LSTM | Time-series prediction model · Ozone | F1-score: 0.284 $\pm$ 0.025 -> 0.302 $\pm$ 0.0155 | Helped |
| Use LTC instead of CT-GRU | Person activity model, 1st setting · Person activity, 1st setting | accuracy: 85.27$\% \pm$ 0.39 -> 85.48$\% \pm$ 0.40 | Helped |
| Use LTC instead of Latent ODE (C enc.) | Person activity model, 2nd setting · Person activity, 2nd setting | accuracy: 0.846 $\pm$ 0.013 -> 0.882 $\pm$ 0.005 | Helped |
| Use LTC instead of LSTM | Half-Cheetah sequence model · Half-Cheetah dynamics | MSE: 2.500$ \pm$ 0.140 -> 2.308$\pm$ 0.015 | Helped |
| Use LTC instead of Neural ODE | Time-continuous network · Computational depth, tanh | computational depth: 0.56 $\pm$ 0.016 -> 9.19 $\pm$ 2.92 | Helped |
| Use LTC instead of Neural ODE | Time-continuous network · Computational depth, sigmoid | computational depth: 0.56 $\pm$ 0.00 -> 7.00 $\pm$ 5.36 | Helped |
Closest papers by meaning
- Identifying nonlinear dynamical systems with multiple time scales and long-range dependencies (ICLR 2021 · similarity 0.88)
- Transformation of ReLU-based recurrent neural networks from discrete-time to continuous-time (ICML 2020 · similarity 0.88)
- AntisymmetricRNN: A Dynamical System View on Recurrent Neural Networks (ICLR 2019 · similarity 0.88)
- Recurrent neural networks: vanishing and exploding gradients are not the end of the story (NeurIPS 2024 · similarity 0.87)
- Signed Graph Neural Ordinary Differential Equation for Modeling Continuous-Time Dynamics (AAAI 2024 · similarity 0.86)
- Almost-Linear RNNs Yield Highly Interpretable Symbolic Codes in Dynamical Systems Reconstruction (NeurIPS 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.