- Rating
- 4(+1)
- Confidence
- 3
- Informativeness
- 3
- Presentation
- 3
- Soundness
- 3(+1)
Read review
Paper summary
The paper studies prototype-based heterogeneous federated learning (HtFL), where clients with different backbone widths exchange class prototypes instead of parameters. It argues that pooling features to a shared dimension only makes prototypes communicable—heterogeneous encoders may express the same class structure under different orthogonal coordinate frames ("gauges"), so direct averaging distorts global prototypes. The authors formalise a geometric aggregatability requirement (common interface + gauge consistency + intrinsic consensus) and propose GeoProtoFL: each client learns an ETF-anchored spherical bi-projection that maps pooled features onto a d-dimensional unit hypersphere; the server registers client prototypes by closed-form orthogonal Procrustes and aggregates them class-wise by a weighted Fréchet barycenter. On HtFLlib (CIFAR-10/100, Tiny-ImageNet; pathological and Dirichlet non-IID) GeoProtoFL achieves the highest mean accuracy against seven HtFL baselines, with single-round theory (Theorems 1–3, Propositions 1–4).
Critical strengths
-
Conceptual contribution. Names and operationalises gauge inconsistency in prototype-based HtFL; reframes ad-hoc prototype refinements (FedTGP / FedORGP / FedSAF) as partial attempts at the same underlying geometric problem. The geometric-aggregatability definition (interface + gauge consistency + intrinsic consensus) is a useful lens for this sub-field.
-
Methodologically tight end-to-end design with no public data. Architecture-agnostic, closed-form server-side operations (SVD-based Procrustes + Karcher iterations for the Fréchet barycenter); communication dimension decoupled from backbone feature dimension (d = 128 / 256 vs dᵢ = 512). Empirically the strongest method in all evaluated main and robustness settings, with a clean ablation (Table 4 V1→V2→V3→V4) whose ordering matches the theory: spherical interface brings the largest single step, but registration only helps when coupled with Fréchet aggregation.
Critical weaknesses
-
The orthogonal Procrustes assumption (Assumption 1) is foundational but is not empirically diagnosed. The backbone continues to update while the bi-projection encoder is frozen, so class directions in the projected space can drift (acknowledged in Remark 1 and the conclusion); the paper does not report how much of the residual mismatch is actually captured by the orthogonal component (e.g. the average ratio F_i/E_i from Theorem 1, or the condition number / rank-deficiency of H_i across the three benchmarks and a few federation scales). Without a measurement, the practical robustness of the closed-form registration to non-orthogonal drift remains an open question.
-
Advantage narrows under sparse participation, and the theory only covers one server round. At N = 100, ρ = 20 % the strong gap to FedORGP collapses to within one standard deviation (37.35 % vs 37.28 %), and Propositions 1–4 + Theorems 1–3 are explicitly per-round with no convergence / generalisation analysis of the federated procedure (App. A.2.5 acknowledges this). The "robust across model heterogeneity and federation scale" claim is therefore carried mostly by the empirical study.
Questions for authors
-
Could you report the average ratio F_i/E_i from Theorem 1 (or the condition number / rank-deficiency of H_i) across the three benchmarks and a few federation scales? If the orthogonal component dominates, that supports Assumption 1 in practice; if not, please discuss how a small non-orthogonal correction (e.g. a low-rank refinement step after Procrustes) would behave.
-
Under sparse participation (ρ = 50 %, 25 %, 20 % in Table 2), could you add a curve of client-frame target stability over rounds (e.g. d_S(\bar p^i_{c,t-1}, \bar p^i_{c,t})) and the staleness of the reused gauge R_i? The conclusion acknowledges that "the advantage narrows under sparse participation"; it would help to know whether this is driven by stale gauges or by reduced per-class coverage.
-
The communication claim is qualitative ("decoupled from dᵢ"). Could you report concrete per-round bytes for GeoProtoFL vs. FedProto / FedTGP / FedORGP / FedSAF on all three benchmarks, with downlink factored in? For small |C_i|, |C_i|\cdot d does not always beat |C_i|\cdot d_i once the head download is included.
12 comments
Continued from X1KG-W1Q1 (1/2).
(3) with known deviations. and cannot be separated in real runs. We therefore treat the registered last-round prototypes as the latent ones, so that is measured, and add a known rotation plus a known anisotropic stretch or per-class noise (levels 0.01–0.2, 10 draws). Selected rows; each column is a median over clients and draws:
| Dataset | Deviation | Error | Bound | Global-prototype displacement, with / without registration | ||
|---|---|---|---|---|---|---|
| CIFAR-10 | stretch 0.1 | 0.128 | 0.012 | 0.154 | 0.00° / 90.1° | |
| CIFAR-10 | noise 0.1 | 0.224 | 0.012 | 0.172 | 0.250 | 4.03° / 87.9° |
| CIFAR-100 | stretch 0.1 | 0.360 | 0.045 | 0.001 | 0.462 | 0.01° / 90.4° |
| CIFAR-100 | noise 0.01 | 0.063 | 0.045 | 0.046 | 0.153 | 0.32° / 92.3° |
| CIFAR-100 | noise 0.1 | 0.627 | 0.045 | 0.585 | 0.734 | 4.58° / 90.2° |
| Tiny-ImageNet | stretch 0.1 | 0.531 | 0.033 | 0.600 | 0.00° / 90.4° | |
| Tiny-ImageNet | noise 0.1 | 0.925 | 0.033 | 0.903 | 0.997 | 4.80° / 89.6° |
Findings.
-
The ratio range is set by the injected levels. ranges from 0.02 to 1.2 (pure rotations, where , excluded). The bound held in all 5,400 cases.
-
Noise tests the bound. Under per-class noise 0.1, the error reaches 69%–91% of the bound, so the bound is informative. Noise is not amplified (error < in every row), and the global prototypes stay within 5° of the unperturbed ones, against about 90° without registration.
-
A stretch is almost entirely absorbed (error ≤ 0.32% of ). This relies on the concentration of the trained prototypes; on spread-out configurations, no rotation can remove a strong stretch.
(4) A non-orthogonal correction. We did not add a non-orthogonal correction, because an unconstrained one would behave badly in two ways:
-
It would copy the reference. With , an unconstrained linear map can send any independent prototypes exactly onto the reference, so registered prototypes would carry no client information. A low-rank refinement fitted on the same prototypes moves toward this solution as its rank grows.
-
It would break the geometry. It would not preserve the geodesic distances used by guidance and aggregation.
After Procrustes, 0.2°–0.5° per class remains (table in (1)). This is small in absolute terms, though not negligible relative to class differences under the practical setting. A refinement regularized toward the Procrustes solution is a natural extension, which we name as future work.
Revision. We have added these measurements to Appendices A.6 and A.7, and one sentence after the orthogonal-gauge model in Section 3.1 and after Assumption 1 (Appendix A.2.1).
W2 & Q2: The advantage narrows under sparse participation (37.35% vs 37.28% at N = 100, ρ = 20%), and the theory covers one server round. Can you show client-frame target stability over rounds and the staleness of the reused gauge, to tell stale gauges from reduced per-class coverage?
Thank you for this question. We agree that the theory covers a single server round and that the margin narrows under sparse participation (Table 2). The measurements below separate the two mechanisms: coverage stays complete, while the global prototypes move more between rounds and reused gauges go stale.
(1) Measurements. CIFAR-100, practical setting (Dirichlet, α = 0.1), 20 participating clients per round as in Table 2, 3 runs, medians over rounds 10–300. The runs reach 43.30% and 37.30%, consistent with Table 2 (43.26% and 37.35%).
A waiting client neither trains nor uploads, so its reused is stale only by how far the global prototypes have moved since it last participated. Stale gauges predict an error that grows with gauge age. Reduced coverage predicts classes without an update, or with fewer contributors.
Measures:
- Target change: angle between a client's targets in consecutive rounds, also for waiting clients.
- Stale-gauge error: angle between the targets from the held gauge and from a gauge refitted on the client's last upload against the current global prototypes.
- Gauge age: rounds since the client last participated.
- Global movement: per-round angle moved by the global prototypes, over the last 100 rounds.
| Setting | Clients per class | Classes without an update | Global movement per round | Target change per round, participating / waiting | Gauge age, median (max) | Stale-gauge error, age 1 → age ≥ 5 |
|---|---|---|---|---|---|---|
| N = 20, ρ = 100% | ≈8 | 0% | 0.0107° | – | 1 (1) | – |
| N = 40, ρ = 50% | 7 | 0% | 0.25° | 0.30° / 0.30° | 1 (16) | 0.13° → 0.21° |
| N = 100, ρ = 20% | 6 | 0% | 0.29° | 0.35° / 0.35° | 4 (52) | 0.17° → 0.35° |
At N = 20, clients per class = 20 × 40 / 100. All angles are small because the prototypes concentrate around a shared direction (R-X1KG-W1Q1). As a scale, the nearest other class is 0.60° (N = 40) and 0.64° (N = 100) from a global class prototype.
(2) Finding.
- Coverage does not change. Each client holds 30–40 of the 100 classes, so 20 participants update every class in the median round; the clients behind each class fall only from about 8 to 6.
- The global prototypes become less stable. With participants changing every round, they move 23× (N = 40) and 27× (N = 100) more per round than with full participation. Targets follow this movement equally for participating and waiting clients, whose gauges are fixed, so the change comes from the global prototypes.
- Reused gauges go stale as a consequence. The stale-gauge error grows with gauge age (Spearman 0.45 / 0.57 at N = 40 / 100). At N = 100 it levels off at about 0.4° after 15 rounds, below the 0.64° to the nearest other class.
- The fit of registration changes little with scale (0.43° → 0.51° → 0.54° per class; R-X1KG-W1Q1).
The narrowing is therefore associated with the movement of the global prototypes between rounds and the stale gauges it causes, not with reduced coverage or with a worse fit of registration.
Revision. We have narrowed the scale claim in Section 4.3.
Thank you for this precise question. On all three benchmarks and at three federation scales, one orthogonal map removes more than 99.99% of the client-to-global distance. Most of this distance lies along a direction shared by a client's classes, which makes Hi nearly rank one, so we test the class structure separately. We answer in four parts: (1) the fit of the orthogonal component, (2) the conditioning of Hi, (3) ∥Fi∥/∥Ei∥ with known deviations, and (4) a non-orthogonal correction. Parts (3) and (4) are in R-X1KG-W1Q1 (2/2).
(1) Fit of the orthogonal component. Practical setting (Dirichlet, α = 0.1), 3 runs, median over clients over the last 100 rounds.
The fit changes little with scale (0.43° → 0.51° → 0.54° per class). What changes is how far the global prototypes move between rounds (R-X1KG-W2Q2).
(2) Conditioning of Hi. By construction, rankHi≤Ci<d. In addition, σ2/σ1 ≤ 1.6×10−4, so Hi is numerically close to rank one: a client's class prototypes and their global counterparts concentrate around a shared direction.