{"id":"bd365d86-1054-40ea-8933-f4d03185155b","arxiv_id":"2504.00890","paper_version":2,"verdict":"UNVERDICTED","confidence":"LOW","novelty_score":7.0,"correctness_risk":"unknown","formal_verification":"none","parameter_count":0,"one_line_summary":"TransNet improves community detection error bounds by adaptively weighting and regularizing eigenspaces from locally privatized source networks with the target network.","lead":"The paper introduces TransNet, a spectral clustering-based transfer learning framework for community detection that uses privacy-preserved auxiliary source networks to improve performance on a target network under local differential privacy. A smart generalist might read it to understand how to leverage multiple networks for better analysis while respecting privacy constraints without needing a trusted curator.","discovery_kind":"new_method","skeptic_critique":{"model":"grok-4.3","headline":"Oracle property requires adaptive weights to perfectly isolate informative sources from perturbed data","rationale":"The reader's weakest_assumption correctly isolates the adaptive weighting step as the load-bearing point. The concern is internal to the argument (whether the weighting construction actually delivers the stated oracle property under local DP) rather than a disagreement with external consensus. Full-text review confirms the claim is stated but does not remove the need for the concrete check above.","tokens_in":1757,"tokens_out":305,"duration_ms":24055,"concrete_test":"Re-derive the weighting formula in the theoretical analysis and apply it to two synthetic source networks (one informative, one non-informative) under randomized response with known privacy parameter ε; compute the aggregated eigenspace error and check whether it equals the oracle bound using only the informative source.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central claim is the error-bound-oracle property: aggregated eigenspace error depends only on informative sources and is no worse than target-only or weighted-sources estimators. This rests on the adaptive weighting scheme (which incorporates privacy budgets and heterogeneity) correctly down-weighting non-informative or heavily privatized sources so their contribution vanishes from the final bound. Because weights are computed from the randomized-response-perturbed adjacency matrices, any distortion in the estimated heterogeneity or privacy-adjusted similarity metrics can leak influence from non-informative sources into the error term, violating the “depends only on informative sources” guarantee.","agreement_with_reader":"agree"},"referee_report":{"model":"grok-4.3","summary":"The manuscript proposes TransNet, a spectral clustering-based transfer learning method for community detection that improves a target network by aggregating eigenspaces from multiple locally stored, heterogeneous source networks under local differential privacy (randomized response perturbation, no trusted curator). An adaptive weighting scheme incorporates privacy budgets and heterogeneity before regularizing the weighted source eigenspace with the target; the central claim is an error-bound-oracle property that the aggregated eigenspace estimation error depends only on informative sources and is no greater than the error of target-only or (weighted) source-only estimators. An extension TransNetX is also presented for the trusted-curator Gaussian perturbation case.","tokens_in":1876,"tokens_out":405,"duration_ms":22879,"significance":"If the oracle property is rigorously established, the work offers a principled, privacy-preserving mechanism for leveraging auxiliary networks in distributed settings where raw data cannot be shared. The guarantee that performance is robust to uninformative or heavily privatized sources, together with the comparison to baseline estimators, would be a meaningful contribution to the intersection of differential privacy and graph transfer learning.","major_comments":[{"comment":"The error-bound-oracle property (abstract and theoretical analysis) is load-bearing for the central claim. The adaptive weighting scheme is computed from the randomized-response-perturbed adjacency matrices; the manuscript must explicitly derive that any estimation error in the privacy-adjusted similarity or heterogeneity metrics cannot allow non-informative sources to contribute to the final error bound, otherwise the “depends only on informative sources” guarantee fails.","section":"theoretical analysis of the adaptive weighting scheme and oracle property"}],"minor_comments":[{"comment":"Notation for the privacy budget and the adaptive weights should be introduced with a single consistent symbol table or definition block to avoid ambiguity when the same quantities appear in both the weighting formula and the error bound.","section":null}],"recommendation":"major_revision","confidential_remarks":null},"author_rebuttal":{"model":"grok-4.3","summary":"We thank the referee for their detailed and constructive review. The concern regarding the robustness of the oracle property under estimation error in the adaptive weights is well-taken, and we address it directly below.","responses":[{"response":"We agree that an explicit derivation is needed to close this gap. While the current analysis establishes the oracle property assuming the weights are computed from the perturbed data, it does not separately bound the effect of randomization on the similarity and heterogeneity metrics used for weighting. In the revision we will add a supporting lemma that (i) quantifies the deviation between the perturbed and unperturbed metrics under randomized response, (ii) shows that this deviation is controlled by the privacy budget and network size, and (iii) demonstrates that any resulting mis-weighting of non-informative sources still keeps their contribution inside the overall error bound (i.e., the final aggregated eigenspace error remains no larger than the target-only or source-only estimators). This addition will be placed immediately before the main oracle-property theorem and will not alter the statement or proof strategy of the existing results.","revision_made":"yes","referee_comment":"[theoretical analysis of the adaptive weighting scheme and oracle property] The error-bound-oracle property (abstract and theoretical analysis) is load-bearing for the central claim. The adaptive weighting scheme is computed from the randomized-response-perturbed adjacency matrices; the manuscript must explicitly derive that any estimation error in the privacy-adjusted similarity or heterogeneity metrics cannot allow non-informative sources to contribute to the final error bound, otherwise the “depends only on informative sources” guarantee fails."}],"tokens_in":1388,"tokens_out":346,"duration_ms":27019,"standing_objections":[]},"desk_editor":{"model":"grok-4.3","letter":"The core advance here is a spectral method that pulls in multiple source networks, each perturbed by randomized response for local DP, then weights their eigenspaces adaptively before regularizing against the target. The claimed payoff is an error bound that depends only on the informative sources and is never worse than target-only or source-only baselines. That oracle property is the main theoretical selling point, and the abstract says the weighting accounts for both privacy budgets and heterogeneity so non-informative or heavily noised sources drop out cleanly. Empirically they report gains across privacy levels and heterogeneity patterns, plus a variant TransNetX that uses Gaussian perturbation when trusted curators exist. The setup is practical for settings where raw edges cannot leave local sites. The adaptive weighting step is genuinely new in this combination of local DP and transfer spectral clustering. The bounds look formally stated and the comparison to simpler estimators is a clean way to show value. The soft spot is exactly where the stress test points: the weights themselves are estimated from the perturbed adjacency matrices, so any error in recovering the heterogeneity or privacy-adjusted similarities could let bad sources leak into the final bound. The abstract does not spell out how the perturbation is controlled inside the weight derivation, which leaves the oracle claim resting on an unverified step. If that step is only shown under strong assumptions or in simulation, the guarantee weakens. This is a methods paper aimed at network analysts who already work with spectral clustering and differential privacy. Readers who need concrete transfer tools for sensitive graph data will find the framework and the empirical comparisons useful even if the bounds need tightening. It is coherent on its own terms and engages the right literature, so it clears the bar for serious refereeing. I would send it out rather than desk reject.","headline":"TransNet adds adaptive eigenspace weighting under local DP to transfer community detection across networks, with an oracle-style error bound that claims to ignore bad sources.","tokens_in":2327,"tokens_out":425,"would_cite":false,"duration_ms":23431,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":{"model":"grok-4.3","evidence":[],"headline":"Privacy-preserving network transfer learning paper has zero overlap with RS forcing chain","alignment":"orthogonal","rationale":"The paper's machinery (adaptive weighting of debiased adjacency eigenspaces under randomized response, error-bound-oracle property for informative sources only, ridge regularization balancing target vs. weighted sources, misclassification bounds under HMSBM) lives entirely in statistical network analysis. It invokes no J-cost, ratio symmetry, φ-ladder, 8-tick periodicity, or parameter-free constant derivations. RS theorems (reality_from_one_distinction, absolute_floor_iff_bare_distinguishability, J-uniqueness via Aczél, D=3 via Alexander duality, etc.) are irrelevant here; the domains are disjoint.","tokens_in":65016,"confidence":"high","tokens_out":166,"duration_ms":16328,"cache_read_input_tokens":32896,"cache_creation_input_tokens":0},"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"grok-4.3","headline":"TransNet achieves an error-bound-oracle property for privacy-preserving community detection by adaptively weighting only informative source eigenspaces.","keywords":["community detection","transfer learning","local differential privacy","spectral clustering","privacy-preserving networks","eigenspace aggregation","randomized response"],"falsifier":"An experiment or counter-example in which adding a non-informative or heavily privatized source increases the final estimation error above the target-only baseline.","tokens_in":2665,"feed_emoji":"🔒","tokens_out":653,"duration_ms":26158,"temperature":0.7,"pith_summary":"The paper introduces TransNet, a spectral clustering framework that improves community detection on a target network by drawing on heterogeneous source networks stored locally and released under local differential privacy via randomized response. It aggregates source eigenspaces with an adaptive weighting scheme that reflects both privacy strength and source quality, then regularizes the result against the target eigenspace. The central theoretical result is that the estimation error of the combined eigenspace depends solely on the informative sources and is guaranteed to be no worse than the error from the target network alone or from the weighted sources alone. A reader would care because the method allows useful transfer without ever moving raw edges and without needing a trusted aggregator, while automatically ignoring useless or over-privatized sources.","feed_headline":"TransNet weights private sources to oracle-bounded community detection","feed_subtitle":"Adaptive eigenspace aggregation depends only on informative networks and matches or beats single-network error bounds under local privacy.","key_machinery":"Adaptive weighting of source eigenspaces under local differential privacy, followed by regularization against the target eigenspace.","core_discovery":"TransNet aggregates source eigenspaces through a novel adaptive weighting scheme that accounts for both privacy and heterogeneity, and then regularizes the weighted source eigenspace with the target eigenspace to optimally balance the two. It establishes an error-bound-oracle property: the estimation error for the aggregated eigenspace depends only on informative sources, ensuring robustness when some sources are highly heterogeneous or heavily privatized. The error bound of TransNet is no greater than that of estimators using only the target network or only weighted sources.","pith_inferences":["The same weighting logic could be tested on other spectral graph tasks such as link prediction under local privacy.","The oracle guarantee suggests the method may compose safely with other federated graph algorithms that also isolate informative participants."],"forward_implications":["Community detection accuracy improves across a range of privacy budgets and source heterogeneity levels.","The procedure stays robust when some sources are useless or over-privatized.","No trusted third party is required because the scheme works entirely in the local differential privacy model.","An extension called TransNetX exists for the case where trusted local curators can apply Gaussian perturbation instead of randomized response."],"fun_headline_variants":["TransNet weights private sources for oracle error bounds","Adaptive private source weighting in TransNet for detection","TransNet error bound depends only on informative sources","Source eigenspaces aggregated with target in TransNet"],"cache_read_input_tokens":64,"weakest_assumption_plain":"The adaptive weighting scheme can effectively account for both privacy levels and heterogeneity in the source networks to achieve the oracle property.","fun_headline_variants_meta":{"raw":{"variants":["TransNet weights private sources for oracle error bounds","Adaptive private source weighting in TransNet for detection","TransNet error bound depends only on informative sources","Source eigenspaces aggregated with target in TransNet"]},"model":"grok-4.3","cost_usd":0.007712,"raw_usage":{"total_tokens":3480,"prompt_tokens":737,"num_sources_used":0,"completion_tokens":58,"cost_in_usd_ticks":77115500,"prompt_tokens_details":{"text_tokens":737,"audio_tokens":0,"image_tokens":0,"cached_tokens":64},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":2685,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":737,"tokens_out":58,"duration_ms":17713,"temperature":1.0,"reasoning_tokens":2685,"cache_read_input_tokens":64,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-05-22T21:50:16.908133+00:00","model_set":{"reader":"grok-4.3"},"falsifier":"An experiment or counter-example in which adding a non-informative or heavily privatized source increases the final estimation error above the target-only baseline.","supporting_citations":[],"review_version":1}