- Rating
- 2
- Confidence
- 2
- Informativeness
- 2
- Presentation
- 2
- Soundness
- 2
Read review
Paper summary
This paper addresses the lack of Lyapunov stability guarantees for online identification of Neural ODEs from streaming data. The theory establishes UUB bounds, improved exponential rates under window persistence of excitation (WPE), and an excitation-free bound on the average functional error. Experiments are conducted on four DeepMind Control Suite domains and two diagnostic systems.
Critical strengths
- Theorem 1 fills the structural gap between the trajectory losses of Neural ODEs and Lyapunov-based adaptive-control analysis.
- The lemmas and propositions support the main results level by level, with all constants given explicitly.
- Appendix verifies a portion of the theoretical results item by item on the actual code and simulated signals, closing the loop between the theory and its empirical verification.
Critical weaknesses
- Footnote 2 and Figure 11(c) acknowledge that the gains were tuned for performance and that the sampled implementations remain stable well below the certified thresholds; however, the paper provides no instance of feasible gains computed from the theorem constants.
- In Table 1, the observer-based baselines fix kf = 200, a value tuned on clean data rather than per noise level, whereas Table 2 shows that CL-LS improves the function error simply by raising kf to 800. The benchmark experiments lack a control in which kf is retuned according to the noise level.
- On clean data, NODE-CL clearly outperforms the same-segment NODE-replay only on pendulum and falls behind on both cartpole and acrobot. The paper offers no explanation or discussion of these results.
Questions for authors
- What distinguishes the proof methodology of this paper from existing Lyapunov stability proof techniques, and in what respects is the derivation novel?
- If X(\tau) in Theorem 1 contains velocity measurement noise, how should the perturbation bound in Eq. (21) be modified? Can the noise term enter the proof in a form that can be absorbed by the sigma-modification gain kσ, so that the theory covers the noisy setting actually evaluated in the experiments?
2 comments
Response to Reviewer L111 (2/2)
Q1: what is new in the proof. The Lyapunov function and the projection and leakage steps are the standard ones. The new element is the object they are applied to. Existing Lyapunov analyses of online adaptation, including concurrent learning and its DNN extensions, start from a gradient that is already linear in the parameter error, a regressor times plus a bounded term, obtained from a derivative, observer, or integral residual in which the model enters at measured states (integral concurrent learning, Parikh et al., 2019, now cited in Section 1, removes the derivative estimate when the model is linear in the parameters). The adjoint gradient of a trajectory loss has no such form; it depends on through a nonlinear rollout. Theorem 1 shows that it nevertheless equals exactly, with built from the mean-value transition matrix of the error dynamics and bounded by constants that depend explicitly on the horizon, through factors of and the Gronwall factor . Three consequences follow: online Neural ODE training over a window is certified (Proposition 3), the horizon becomes a computable condition instead of a modeling assumption, and the Gauss-Newton matrix assembled from forward sensitivities is shown to approximate the exact operator with a quantified gap (Lemma 11), which is what brings a second-order gain inside the certificate. The revised introduction states this.
Q2: velocity measurement noise in Theorem 1. Yes, for bounded noise, and in the form you anticipate. Appendix I (Proposition 5) takes the measured trajectory , , with the rollout started from the measured state and the loss evaluated against the measured trajectory, as implemented. The decomposition holds with constructed as in Theorem 1, from the true trajectory and the rollout started at the measured state, and Eq. (21) becomes
with unchanged, under the window condition . Proposition 3 and Theorem 2 hold for bounded noise with (and of Lemma 11) replaced by these constants. The gain condition of Proposition 3 changes by a term linear in (that of Theorem 2 also through , quadratic in ), the residual changes through , and absorbs the noise exactly as it absorbs the approximation error. The proposition gives the mechanism behind the noise results of Table 1: in the observer pathway the noise is differentiated and enters with a factor that grows with the observer gain; in the trajectory pathway it is integrated over the segment and enters with the factor , close to for short segments.
Response to Reviewer L111 (1/2)
Thank you for the review. The three weaknesses and two questions in order.
W1: an instance of feasible gains computed from the theorem constants. Provided in Part B of the general response and in the new Appendix L.5. On S1 seed 0, with closed-form upper bounds on the derivative constants of the network over Ω×Θϵ, margin ϖ=1 and segment length Ts=2 ms, the containment condition Eq. (17) holds and Eq. (25) at α=2 reads kσ>0.076. The run with (α,kσ)=(2,0.084) and Γ∈[10,100] satisfies the hypotheses of Theorem 2 and is stable; the leakage holds the estimate at the origin, which is why such gains are not the ones used. At the tuned segment length of 50 ms the same condition requires a far larger kσ (1.3×105 with constants maximized over sampled parameters), against 1.2×10−2 when the constants and the error bound are taken along the realized run, and 10−9 in the experiments. Part B explains why a gap of that size is expected: a worst case over a parameter set chosen large, slope constants and error bounds that multiply, and norm bounds that assume adversarial alignment. The condition is sufficient in the same sense as a step-size bound η≤1/L in optimization theory, the certified baselines share the property, and Part B lists what holds at the tuned gains and what Theorem 2 adds.
W2: kf retuned per noise level. Done (Appendix L.2, Table 7, Section 4.1): CL and CL-LS at σv=3×10−3 with kf=αo∈{50,100,200,400,800}, four domains, five seeds, all else unchanged. All entries are from one machine (Table 4 reruns Table 1 there).
Retuning helps the observer-based methods, and NODE-CL remains ahead of retuned CL-LS by factors of 13, 2.7, 1.3, 1.8 and of retuned CL by 8.4, 3.0, 1.5, 13. The direction is the one the observer analysis gives: the continuous-time floor zˉo falls as kf grows, while the sampled observer differentiates the velocity noise with a factor that grows with kf, so noise moves the best kf downward (to 200 or less on every domain) and the error rises steeply above 200 on the cartpole and reacher; the floor term limits how far lowering kf can help. The improvement at kf=800 in Table 2 is on the noise-free diagnostic runs, where only the floor term is present. NODE-CL has no observer and no gain that must track the noise level. On the pendulum with twenty seeds (Table 5 and Appendix L.1) the best medians on the same grid are 0.572 for CL-LS and 0.355 for CL, against 0.099 for NODE-CL at its Table 1 setting.
W3: clean-data results against NODE-replay. Two points, both in the revision.