- Rating
- 2
- Confidence
- 3
- Informativeness
- 3
- Presentation
- 2
- Soundness
- 2
Read review
Paper summary
UniCOP separates posterior sample construction from the inference mechanism. A Gaussian latent matrix is skew-symmetrized and transformed into a Cayley rotation of a learned, non-sampled reference weight. Samples preserve the reference row Gram matrix, singular values, and Frobenius norm. The paper develops a conditional latent-space ELBO, geometric characterization, and function-space uncertainty analysis, then evaluates several BNN inference methods and feature-posterior transfer to CLAP4CLIP.
Critical strengths
-
The probability-space formulation appropriately handles orbit-supported distributions. The latent ELBO and KL contraction handle singular weight distributions correctly. The exact structural invariants and local tangent coverage establish the geometry of the resulting support. The ELBO is appropriately conditional on the learned, non-sampled reference. [pp. 4–5; Appendix A.1.1–A.1.2, pp. 13–15]
-
Practical evidence extends beyond the disputed mechanistic control. Independently trained IVON+UniCOP improves all four reported mean metrics across CIFAR-100, Tiny-ImageNet, and ImageNet-100, including accuracy gains of 0.36, 0.48, and 1.12 percentage points. On held-out CLAP4CLIP class orders, future-class MI AUROC improves by 1.73 points, with a 0.54-point final-accuracy reduction. [Table 8, p. 22; Table 10, p. 24]
Critical weaknesses
- The Euclidean control does not generally match local perturbation energy. Equations (59)–(60) use a dimension-only scale, whereas the Cayley Jacobian depends on the latent center and its expected energy depends on the covariance. Writing for the constructor, including skew extraction, take , reference row , skew center with and , and iid standard-normal perturbation entries in ,
while the Euclidean derivative has expected squared norm . At , the comparator therefore has fourfold local energy. The comparison also decodes a Cayley-trained posterior through an alternative map, including directions discarded by skew extraction. It therefore does not isolate geometry at matched perturbation energy. [Eq. (33), pp. 14–15; Eqs. (59)–(60), Table 7, p. 20]
- The practical cost–benefit comparison remains underspecified. The BNN layers, convolution grouping, blocks, and reference/latent update rules are insufficiently specified. A dense -dimensional Cayley solve requires arithmetic and latent storage; blocking changes both cost and accessible geometry. Appendix A.4 specifies CLAP4CLIP blocking, while also disclosing unmatched posterior parameter counts. Without analogous BNN details and measured overhead, the accuracy/calibration/OOD trade-offs cannot be assessed against resource-matched alternatives. [p. 4; Appendices A.2 and A.4–A.6, pp. 18–24]
Questions for authors
-
What are the actual Jacobian-energy ratios at the trained checkpoints, and do Table 7’s ranking gains survive correct energy matching? A corrected control is needed to attribute the difference to posterior geometry.
-
Please specify the BNN layer/block implementation, parameter counts, reference/latent updates, and measured training/inference time and memory.
-
Under matched parameter and sampling budgets, how do the OOD gains compare with accuracy and calibration costs, particularly in CLAP4CLIP? Corrected attribution and a favorable, documented resource–quality trade-off would support a rating of 3.