- Rating
- 2
- Confidence
- 2
- Informativeness
- 2
- Presentation
- 3
- Soundness
- 2
Read review
Paper summary
The paper proposes IsoAct, a post-hoc debiasing framework for representations on known manifolds, including the hypersphere, Lorentz hyperboloid, and SE(3). It identifies nuisance-predictive directions using Fisher Discriminant Analysis (FDA) in tangent coordinates, constructs a neutral reference from low-confidence training examples, and applies manifold-native updates to reduce nuisance predictability while preserving manifold membership. Experiments on synthetic benchmarks, KLIFS ligand poses, and CLIP/MERU embeddings evaluate nuisance reduction, downstream utility, and geometric validity.
Critical strengths
-
Debiasing manifold-valued representations is a meaningful and relatively underexplored problem. Jointly evaluating nuisance reduction, downstream utility, and manifold validity provides a useful perspective.
-
The method offers a coherent post-hoc pipeline across three distinct geometries, with reasonably detailed mathematical formulations and implementation specifications.
-
The empirical results are encouraging in several settings, particularly the consistent debiasing improvements on MERU and the high retrieval retention on CLIP.
Critical weaknesses
-
IsoAct provides no guarantee of linear nuisance erasure. Matching projections along FDA directions does not necessarily equalize class means, so the nuisance can remain recoverable by linear probes, unlike LEACE’s guaranteed linear erasure in Euclidean space.
-
The real-world experiments do not isolate the benefits of manifold-native editing: KLIFS selects translation-only updates, all MERU baselines are already manifold-valid, and CLIP lacks comparisons with renormalized baselines. Natural competitors such as Euclidean editing with projection, log–LEACE–exp, and WRING are also missing.
Questions for authors
-
How does IsoAct compare with Euclidean editing followed by manifold projection, or with a log–LEACE–exp baseline?
-
Why does the selection rule pick translation-only editing on KLIFS, and how much kinase-group information remains in the unedited rotations (e.g., from a probe on rotations alone)?
4 comments
2. Kinase-group information in rotation and translation component (Q2/W2). To examine the bias information in each component, we used the same split and train-fitted linear probe:
R (9D), t (3D), and R+t (12D) denote the flattened rotation, translation, and their concatenation, respectively. The same nuisance metrics and bootstrap SD protocol apply.
| Features | N.Acc±SD | N.F1±SD |
|---|---|---|
| Null | 0.0694±0.0022 | 0.0603±0.0021 |
| Original R (9D) | 0.1036±0.0061 | 0.0879±0.0054 |
| Original t (3D) | 0.1560±0.0072 | 0.1486±0.0067 |
| Original R+t (12D) | 0.1398±0.0067 | 0.1409±0.0067 |
Although the rotation component also contains probe-readable kinase-group information, translation supports stronger prediction. This provides an empirical explanation for why translation-only editing was selected in this setting.
What's more, we would like to clarify that the selection rule did not specify the type of IsoAct configurations. The selection rule balances nuisance reduction and utility across available actions, and the translation-only editing was chosen among candidate pool.
3. Linear nuisance erasure (W1). We agree that IsoAct does not provide LEACE's guarantee of linear nuisance erasure. Matching projections along FDA directions does not by itself guarantee equal nuisance-class means, so linear predictability may remain. IsoAct aims to reduce nuisance predictability while retaining utility and manifold membership. We therefore added LEACE and its projected variant to compare these trade-offs empirically, as shown above.
We will add these comparisons and component-wise probes to the revision to clarify both the effect of projection and the selection of translation-only editing.
I appreciate the authors’ additional experiments and clarification. The component-wise probes explain the translation-only choice on KLIFS, and the projection-repair comparisons strengthen the evaluation. However, several key concerns remain: The CLIP experiments still lack SPD with L2 normalization and WRING, which is directly applicable to this setting. For KLIFS, the main result still relies on translation-only editing. The joint-editing experiment demonstrates feasibility but does not isolate the contribution of SO(3) editing. So the experiments do not establish a clear advantage over the simplest or closely related alternatives. Therefore, I will maintain my score.
Thank you for recognizing that the component-wise probes explain the translation-only choice and that the projection-repair comparisons strengthen the evaluation. We present further CLIP comparisons and a controlled KLIFS ablation to address your remaining concerns.
1. SPD with L2 normalization and WRING on CLIP.
IsoAct achieves higher mean retrieval retention than SPD+Norm and WRING across all three attributes, with the highest for age.
Under the same setting as our reported CLIP results, we added SPD followed by L2 normalization (SPD+Norm) and WRING. Following Table 3 of the revised manuscript, averages
over protected-attribute Accuracy and macro-F1. Higher indicates greater proximity to Null, the shuffled-label reference. Retrieval retention averages the method/Original ratios for R@1, R@5, and R@10. The constraint column reports mean unit-norm error. Bold marks the highest and retention within each attribute among the reported methods.
| Attribute | Method | Mean ↑ | Mean retrieval retention ↑ | Mean unit-norm error ↓ |
|---|---|---|---|---|
| Age | SPD | 0.7536 | 0.9867 | 1.8327e-02 |
| Age | SPD+Norm | 0.8111 | 0.9868 | 3.3660e-09 |
| Age | WRING | 0.7984 | 0.9881 | 2.2936e-08 |
| Age | IsoAct | 0.9234 | 0.9890 | 6.3230e-08 |
| Gender | SPD | 0.8243 | 0.9938 | 1.0768e-02 |
| Gender | SPD+Norm | 0.4289 | 0.9939 | 3.1858e-09 |
| Gender | WRING | 0.3704 | 0.9990 | 2.3162e-08 |
| Gender | IsoAct | 0.4254 | 1.0003 | 3.5303e-08 |
| Race | SPD | 0.9656 | 0.9891 | 5.3183e-02 |
| Race | SPD+Norm | 0.9324 | 0.9907 | 3.5094e-09 |
| Race | WRING | 0.8810 | 0.9926 | 2.2442e-08 |
| Race | IsoAct | 0.8022 | 0.9932 | 6.1190e-08 |
Age. Normalization improves SPD’s debiasing, while IsoAct achieves the highest and retrieval retention among the compared methods.
Gender. IsoAct exceeds WRING on both metrics and achieves higher retrieval retention than SPD+Norm at a similar . Normalization reduces SPD’s from 0.8243 to 0.4289 despite restoring near-zero unit-norm error.
Race. SPD+Norm and WRING achieve higher , while IsoAct achieves higher mean retrieval retention.
These comparisons address the simpler normalization baseline and directly applicable WRING method you identified as missing. Since SPD+Norm, WRING, and IsoAct all achieve near-zero unit-norm error, these results strengthen the empirical case for IsoAct beyond satisfying the output constraint.
2. Isolating the contribution of rotational editing on KLIFS.
To address your concern that the joint-editing experiment "does not isolate the contribution of SO(3) editing," we conducted a controlled ablation.
Starting from the joint configuration in our previous response, we restored the original rotation of each edited pose to obtain . The two conditions differ only in rotation; translation outputs are exactly identical, and all other experimental settings are unchanged.
N.Acc and N.macro-F1 measure kinase-group prediction. Both conditions remain above Null, so lower nuisance scores are better here; higher task scores are better. Entries are full-test estimates ± SD from 1,000 paired test bootstrap resamples. Bold marks the better point estimate.
| Configuration | N.Acc | N.macro-F1 | aC F1 | aC Recall | aC AUROC | Pocket F1 | Pocket Recall | Pocket AUROC |
|---|---|---|---|---|---|---|---|---|
| Original rotations restored | 0.1314 ± 0.0066 | 0.1302 ± 0.0066 | 0.3907 ± 0.0251 | 0.3391 ± 0.0255 | 0.7262 ± 0.0148 | 0.8355 ± 0.0115 | 0.8203 ± 0.0114 | 0.9717 ± 0.0021 |
| Joint editing | 0.1283 ± 0.0067 | 0.1167 ± 0.0063 | 0.3922 ± 0.0248 | 0.3448 ± 0.0253 | 0.7382 ± 0.0141 | 0.8288 ± 0.0129 | 0.8241 ± 0.0122 | 0.9722 ± 0.0022 |
Keeping the edited rotations brings both nuisance scores closer to Null, with macro-F1 decreasing from 0.1302 to 0.1167. It also yields higher point estimates for all three aC metrics and Pocket Recall/AUROC; Pocket F1 is lower (0.8288 versus 0.8355).
Thus, this controlled comparison isolates a contribution of SO(3) editing to nuisance reduction and downstream performance with translation held fixed.
We sincerely appreciate the time and attention you have devoted to our paper throughout the discussion. Your feedback has helped us strengthen the empirical evaluation and clarify the scope and contributions of our work.
Thank you for recognizing the importance of manifold-valued debiasing, the coherent framework across three geometries, and the encouraging MERU and CLIP results. Your suggestions on projection baselines and component-wise probes guided the additional KLIFS experiments below.
1. Euclidean editing followed by projection (Q1/W2). By evaluating the performance of 'projected' variants of baselines in KLIFS setting, we found that projection restores manifold constraints, but its effects on debiasing and utility are not coherent.
The details of the projection are as follow: we project each selected output rotation onto SO(3), with determinant correction. Raw/projected pairs share the debiasing configuration. Entries of the table are full-test estimates ± SD from 1,000 paired test bootstrap resamples with fitted models fixed. C1/C2 are mean orthogonality/determinant errors.
Also, we have added LEACE as a new baseline to compare their empirical debiasing–utility trade-offs. While your suggestion of 'log–LEACE–exp' method is also valid, we chose to apply the same post-projection repair protocol to all baselines for consistency.
N.Acc and N.F1 denote kinase-group prediction accuracy and macro-F1. aC and Pocket are the two downstream tasks; their F1 and Rec scores concern the minority class, and AUC denotes AUROC. Null is the shuffled-label nuisance reference: proximity to Null indicates stronger debiasing, while higher utility scores are better. C1/C2 measure rotation orthogonality/determinant errors, for which lower is better.
The results show why geometric validity alone is insufficient. Projection has mixed effects on debiasing and utility. For SPD, projection repair worsens debiasing while showing mixed impact on utility preservation. For LEACE, projection brings nuisance Acc closer to Null, yet decreases five of six utility metrics. IsoAct combines reduced nuisance predictability with the highest aC F1, Recall, and AUROC in this comparison and near-zero constraint errors. Projection repair alone does not reproduce this trade-off.