{"id":"44016970-a8e0-43c3-84c5-263043048121","arxiv_id":"2604.07298","paper_version":1,"verdict":"UNVERDICTED","confidence":"LOW","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"ROAM routes region tokens to MoE experts via entropic optimal transport with per-slide capacity marginals and graph regularization, achieving competitive performance and external AUC 0.845 on NSCLC WSI benchmarks.","lead":"ROAM introduces a mixture-of-experts aggregator for whole-slide pathology images that routes spatially binned region tokens to experts using capacity-constrained optimal transport on a region graph. This aims to prevent expert collapse and better handle tissue heterogeneity than standard MIL or unconstrained MoE methods.","discovery_kind":"new_method","skeptic_critique":{"model":"grok-4.3","headline":"Graph-regularised Sinkhorn may violate the per-slide capacity marginals that are claimed to enforce balanced expert utilisation by construction.","rationale":"The reader's weakest assumption correctly flags the graph regularisation and information preservation, but the more load-bearing risk is whether the regularisation preserves the marginal constraints that justify the entire MoE construction. The performance numbers themselves are not directly challenged; only the attribution to the proposed balancing mechanism is.","tokens_in":1786,"tokens_out":332,"duration_ms":50569,"concrete_test":"Extract the exact Sinkhorn update rule and graph regularisation operator from §3 (or the supplementary algorithm box). Re-run the forward pass on one TCGA-CPTAC slide using the published region graph and capacity marginals; compute the L1 deviation of the obtained expert marginals from the target marginals both with and without the graph term. If the deviation exceeds 0.05 in the regularised case, the balance guarantee does not hold.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central technical claim is that capacity-constrained entropic OT (Sinkhorn) with explicit per-slide marginals guarantees balanced routing without auxiliary losses, while graph-regularised iterations only diffuse assignments over the region graph. If the regularisation is implemented by modifying the cost matrix or by post-hoc diffusion steps that are not re-projected onto the marginals, the resulting transport plan can deviate from the prescribed uniform (or capacity) marginals. This directly undermines the stated advantage over unconstrained softmax routing and makes the performance competitiveness rest on an unverified invariance rather than on the stated mechanism.","agreement_with_reader":"partial"},"referee_report":{"model":"grok-4.3","summary":"The manuscript introduces ROAM, a spatially aware MoE-MIL aggregator for gigapixel WSI classification. Patch embeddings are compressed into spatially binned region tokens; these are routed to a pool of expert poolers via capacity-constrained entropic optimal transport (Sinkhorn) whose per-slide marginals are intended to enforce balanced expert utilisation by construction. Graph-regularised Sinkhorn iterations are added to diffuse assignments over the region adjacency graph. The paper reports competitive performance against MIL and MoE baselines on four WSI benchmarks and an external AUC of 0.845 ± 0.019 on TCGA-CPTAC NSCLC generalisation.","tokens_in":1935,"tokens_out":447,"duration_ms":98803,"significance":"If the routing mechanism truly preserves the prescribed marginals while adding spatial coherence, ROAM supplies a principled, auxiliary-loss-free alternative to unconstrained softmax routing in pathology MoE models. The explicit use of capacity marginals and the external validation set are concrete strengths that would support broader adoption of OT-based routing in computational pathology.","major_comments":[{"comment":"The central claim that balanced expert utilisation is guaranteed 'by construction' rests on the transport plan satisfying the per-slide capacity marginals after graph regularisation. The manuscript must demonstrate (via pseudocode, convergence argument, or explicit reprojection step) that the graph-diffusion operation does not violate these marginals; otherwise the stated advantage over standard softmax routing is not established.","section":"Method (graph-regularised Sinkhorn iterations)"}],"minor_comments":[{"comment":"The abstract states 'competitive results' and an external AUC but supplies neither the exact baseline implementations, number of experts, nor statistical tests; these must be detailed in the results section with tables showing per-expert utilisation statistics.","section":"Abstract and Results"},{"comment":"Clarify whether the reported ±0.019 on the external AUC is standard deviation across runs, cross-validation folds, or bootstrap; add this to the evaluation protocol description.","section":"Results (external validation)"}],"recommendation":"major_revision","confidential_remarks":null},"author_rebuttal":{"model":"grok-4.3","summary":"Thank you for the thorough review of our manuscript. We respond to the major comment point-by-point below and will incorporate the necessary clarifications in the revised version.","responses":[{"response":"We thank the referee for highlighting this important point. The graph-regularised Sinkhorn procedure is constructed so that diffusion occurs as a convex combination within each iteration, after which the standard Sinkhorn row/column scaling steps are applied to restore the prescribed marginals. In the revision we will add explicit pseudocode (new Algorithm 1) together with a short convergence argument showing that the final transport plan satisfies the per-slide capacity marginals upon termination. This will rigorously establish the claimed advantage over unconstrained softmax routing. We will also include empirical expert-utilisation histograms on the validation sets to corroborate the theoretical guarantee.","revision_made":"yes","referee_comment":"[Method (graph-regularised Sinkhorn iterations)] The central claim that balanced expert utilisation is guaranteed 'by construction' rests on the transport plan satisfying the per-slide capacity marginals after graph regularisation. The manuscript must demonstrate (via pseudocode, convergence argument, or explicit reprojection step) that the graph-diffusion operation does not violate these marginals; otherwise the stated advantage over standard softmax routing is not established."}],"tokens_in":1414,"tokens_out":278,"duration_ms":49720,"standing_objections":[]},"desk_editor":{"model":"grok-4.3","letter":"The core contribution is a region-graph OT router that assigns spatially binned tokens to experts using Sinkhorn with explicit capacity marginals per slide, plus graph-regularized iterations to encourage local coherence. This is a direct attempt to enforce balanced utilisation without the usual auxiliary losses, and the abstract reports competitive numbers plus a solid external AUC of 0.845 on TCGA-CPTAC NSCLC data. That external result is the strongest piece of evidence they provide for generalisation under frozen foundation-model embeddings. The framing of the problem—standard MIL and unconstrained MoE both collapse under slide heterogeneity—is accurate and the proposed fix stays within standard OT machinery, which is a plus for reproducibility. The region token construction also aligns routing with tissue neighbourhoods, which makes sense for pathology. On the downside, the abstract and available details give no ablations that isolate the OT marginals from the graph regularisation or from the binning step itself. Without those, it is difficult to know whether the reported gains trace to the claimed mechanism. The stress-test concern about marginal violation is real: if the graph diffusion is applied by modifying the cost matrix or by post-hoc steps that are not re-projected onto the prescribed marginals, the transport plan can drift and the “balanced by construction” claim weakens. The paper would need to show the final row and column sums stay close to the targets after regularisation. Implementation specifics on how the Sinkhorn iterations are altered are also thin. This work is aimed at people already doing MIL or MoE in computational pathology who want a spatially aware aggregator. It is not broad enough or novel enough to change practice outside that niche, but the technical idea is coherent enough that a serious referee should see it. I would send it to review with a request for the missing ablations and a marginal-preservation check.","headline":"ROAM gives a clean OT-based fix for expert collapse in MoE-MIL on WSIs via per-slide marginals and graph diffusion, but the regularization step risks breaking the balance guarantee and the experiments lack the controls needed to confirm the mechanism works.","tokens_in":2441,"tokens_out":458,"would_cite":false,"duration_ms":34338,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":{"model":"grok-4.3","evidence":[{"relation":"unclear","rs_module":"IndisputableMonolith/Cost/FunctionalEquation.lean","rs_theorem":"washburn_uniqueness_aczel","paper_passage":"region-to-expert assignment formulated as entropic optimal transport (Sinkhorn) with explicit per-slide capacity marginals, enforcing balanced expert utilisation without auxiliary load-balancing losses"},{"relation":"unclear","rs_module":"IndisputableMonolith/Foundation/AlexanderDuality.lean","rs_theorem":"alexander_duality_circle_linking","paper_passage":"graph-regularised Sinkhorn iterations that diffuse routing assignments over the spatial region graph"}],"headline":"ROAM Sinkhorn routing with capacity marginals and graph diffusion has no structural overlap with RS J-cost forcing, ratio symmetry or distinction-derived constants.","alignment":"orthogonal","rationale":"Paper's core machinery (entropic OT with explicit marginals q=1/E, graph-regularised Sinkhorn iterations, cosine-dissimilarity costs on region tokens) is standard ML optimal-transport routing for MoE load balancing. RS framework derives J(x)=½(x+x⁻¹)−1, φ-ladders, 8-tick periodicity and parameter-free constants from a single distinction; none of these appear in the paper's formulation, proofs or ablations. No echo of recognition-cost functional equations, Alexander duality for D=3, or cosh-cost identities is present.","tokens_in":46192,"confidence":"low","tokens_out":334,"duration_ms":12894,"cache_read_input_tokens":128,"cache_creation_input_tokens":0},"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"grok-4.3","headline":"ROAM uses capacity-constrained optimal transport and graph regularization to balance expert routing for whole-slide image classification.","keywords":["whole-slide image classification","mixture of experts","optimal transport","multiple instance learning","computational pathology","graph regularization","entropic optimal transport"],"falsifier":"Removing the capacity marginals or the graph-regularization term produces either markedly imbalanced expert utilization or lower AUC on the same four WSI benchmarks and external NSCLC set.","tokens_in":2699,"feed_emoji":"🔬","tokens_out":618,"duration_ms":56975,"temperature":0.7,"pith_summary":"Standard multiple-instance learning routes every patch through the same pathway, while unconstrained mixture-of-experts models often collapse to one or two dominant experts. ROAM first compresses patches into spatial region tokens that respect tissue neighborhoods, then solves region-to-expert assignment with entropic optimal transport under explicit capacity marginals per slide. Sinkhorn iterations are further regularized by diffusion across the region graph so that neighboring regions tend to select the same experts. The result is balanced utilization enforced by construction rather than by auxiliary losses. The method is evaluated on four WSI benchmarks with frozen embeddings and reports competitive accuracy plus an external AUC of 0.845 on TCGA-CPTAC NSCLC slides.","feed_headline":"Optimal transport balances experts for slide-image classification","feed_subtitle":"ROAM compresses patches to region tokens and uses Sinkhorn with graph regularization to avoid expert collapse, reaching 0.845 external AUC.","key_machinery":"Graph-regularised Sinkhorn iterations for capacity-constrained entropic optimal transport on spatial region tokens.","core_discovery":"ROAM formulates region-to-expert assignment as entropic optimal transport with explicit per-slide capacity marginals solved by Sinkhorn iterations, with additional graph regularization over the spatial region graph, thereby enforcing balanced expert utilization by construction while aligning routing with local tissue neighborhoods.","pith_inferences":["The same optimal-transport routing could be tested on other spatially structured MIL tasks such as remote-sensing scene classification.","Graph diffusion may implicitly encode larger-scale pathological patterns that single-region decisions miss.","Entropic optimal transport offers a differentiable alternative to softmax routing that may scale to larger expert pools."],"forward_implications":["Balanced expert utilisation occurs without any auxiliary load-balancing losses.","Routing decisions respect local tissue neighborhoods through spatial binning and graph diffusion.","Performance remains competitive with strong MIL and MoE baselines across four WSI benchmarks.","External generalisation reaches 0.845 AUC on TCGA-CPTAC NSCLC data.","Region-token compression reduces the number of routing decisions while retaining neighbourhood structure."],"fun_headline_variants":["Region-graph OT routes tokens to balanced MoE experts","Capacity Sinkhorn balances experts in pathology MoE","Graph-regularized transport for region-expert routing","Entropic OT enforces expert balance via region graphs"],"cache_read_input_tokens":2112,"weakest_assumption_plain":"Compressing patches into spatially binned region tokens plus graph regularization preserves enough instance-level information to justify the added routing machinery and to improve downstream classification.","fun_headline_variants_meta":{"raw":{"variants":["Region-graph OT routes tokens to balanced MoE experts","Capacity Sinkhorn balances experts in pathology MoE","Graph-regularized transport for region-expert routing","Entropic OT enforces expert balance via region graphs"]},"model":"grok-4.3","cost_usd":0.005948,"raw_usage":{"total_tokens":2767,"prompt_tokens":722,"num_sources_used":0,"completion_tokens":58,"cost_in_usd_ticks":59478000,"prompt_tokens_details":{"text_tokens":722,"audio_tokens":0,"image_tokens":0,"cached_tokens":64},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":1987,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":722,"tokens_out":58,"duration_ms":34460,"temperature":1.0,"reasoning_tokens":1987,"cache_read_input_tokens":64,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-05-10T18:14:45.436005+00:00","model_set":{"reader":"grok-4.3"},"falsifier":"Removing the capacity marginals or the graph-regularization term produces either markedly imbalanced expert utilization or lower AUC on the same four WSI benchmarks and external NSCLC set.","supporting_citations":[],"review_version":1}