On Second Order Behaviour in Augmented Neural ODEs
Authors: Alexander Norcliffe, Cristian Bodnar, Ben Day, Nikola Simidjievski, Pietro Lió
NeurIPS 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Finally, we compare SONODEs and ANODEs on synthetic and real dynamical systems and demonstrate that the inductive biases of the former generally result in faster training and better performance. |
| Researcher Affiliation | Academia | Alexander Norcliffe Department of Physics University of Cambridge alex.norcliffe98@gmail.com Cristian Bodnar , Ben Day , Nikola Simidjievski, Pietro Liò Department of Computer Science and Technology University of Cambridge {cb2015, bjd39, ns779, pl219}@cam.ac.uk |
| Pseudocode | No | No section or figure explicitly labeled 'Algorithm' or 'Pseudocode' was found, nor were there any structured code-like blocks. |
| Open Source Code | Yes | Our code is available online at https://github.com/ a-norcliffe/sonode. |
| Open Datasets | Yes | one-dimensional compact parity experiment (originally named g1d in Dupont et al. [4]), nested n-spheres problem [4], airplane vibrations [16], The Silverbox dataset [21] |
| Dataset Splits | No | The paper mentions 'training points' and 'test points' and 'trained on the first 1000 timestamps and then extrapolated to the next 4000', but it does not explicitly specify a validation dataset split or provide details on how a validation set was used. |
| Hardware Specification | No | The paper does not specify any particular hardware components like CPU or GPU models, processor types, or memory amounts used for running the experiments. |
| Software Dependencies | No | The paper does not provide specific ancillary software details, such as library names with version numbers (e.g., 'PyTorch 1.9' or 'Python 3.8'). |
| Experiment Setup | Yes | For such tasks, we use the mean squared error (MSE) between these points and the corresponding predicted location over all time steps for training the models. For the few toy classification tasks we include, we optimise only for the linear separability of the final positions via the cross-entropy loss function., These were each evolved for 10 seconds, using one hundred evenly spaced time stamps., The models were trained on fifty training points in the first ten seconds of x = sin(t), and then tested with ten points in the next five seconds. The train points had noise added to them, drawn from a normal distribution N(0, σ2) for different standard deviations σ = (0, 0.1, 0.2, . . . , 0.7). |