- Rating
- 2
- Confidence
- 2
- Informativeness
- 2
- Presentation
- 2
- Soundness
- 2
Read review
Paper summary
This paper studies inexact high-order proximal-point methods (Inexact HiPPA) for nonsmooth, nonconvex objectives motivated by robust representation alignment. The authors establish strong quasar-convexity for several stabilized capped and anisotropic losses under clean realizability and identifiability assumptions. They prove global convergence under summable model-value inexactness and, assuming additional local stability and accuracy conditions, linear convergence for and superlinear convergence of Q-order for . They also derive stability and local recovery results under target corruption. Experiments on CIFAR-100 features examine local convergence rates and robust alignment under controlled replacement corruption.
Critical strengths
-
Meaningful geometric characterization of robust objectives. The paper establishes strong quasar-convexity for several nonsmooth, nonconvex capped and anisotropic alignment losses, with explicit constants and identifiable conditions.
-
Careful theoretical formulation and controlled validation. The paper distinguishes three notions of proximal inexactness, provides verifiable local convergence examples, and reports lower-level computational work alongside outer iterations.
Critical weaknesses
-
Restricted applicability and unclear incremental significance of the main convergence results. Clean realizability makes the ideal alignment map directly recoverable by least squares, whereas corruption removes the global quasar-convex guarantee. The Q-order result assumes a corresponding local recurrence (Assumption 10), explicitly verified only in a locally quadratic basin; meanwhile, the motivating nonsmooth examples exhibit finite proximal identification for every (Remark 31). This limits the demonstrated generality and leaves the novelty relative to existing HiPPA theory insufficiently established.
-
Insufficient evidence for practical computational and robustness advantages. The local-rate experiment uses oracle-dependent accuracy schedules and remains inside a quadratic basin, while the corrupted experiment compares different orders using a common without demonstrating comparable regularization scales or tuned computational budgets. The striking versus endpoint result therefore requires further validation through matched evaluations, wall-clock time, and stronger optimizer baselines. Moreover, the quadratic stabilizer yields unbounded large-residual influence, leaving the robustness advantage over alternative robust losses insufficiently characterized.
Questions for authors
- Can Assumption 10(b) be verified for a meaningful non-quadratic, nonsmooth objective without finite proximal identification? What are the precise new rate guarantees relative to Ahookhosh et al. (2026a,b) and Ahookhosh and Nesterov (2024)?
- Can the claimed rates be retained using a fully computable stopping and forcing criterion that requires neither the unknown proximal minimum nor the distance to ? What rate guarantees apply to the practical residual-and-descent rule?
- How do the versus results change with separately tuned or scale-covariant , comparable accuracy schedules, and matched wall-clock or lower-level computational budgets? Could the authors also report test MSE and variability?