{"id":"22c51c1e-815e-45e6-88a7-9d35537be38a","arxiv_id":"2606.31473","paper_version":1,"verdict":"UNVERDICTED","confidence":"LOW","novelty_score":3.0,"correctness_risk":"unknown","formal_verification":"none","parameter_count":0,"one_line_summary":"Von Mises ensemble yields angular uncertainty estimates that integrate directly into tracking via closed-form likelihoods and shows stronger perturbation sensitivity than evidential deep learning on radar DOA tasks.","lead":"This paper compares a von Mises ensemble approach with evidential deep learning for modeling uncertainty in direction-of-arrival estimates from automotive radar. A smart generalist might read it to see how circular statistics could help radar systems in self-driving cars handle ambiguous targets more reliably.","discovery_kind":"new_application","skeptic_critique":{"model":"grok-4.3","headline":"Integration claim asserted via closed-form VM likelihoods but downstream tracking pipeline never tested","rationale":"Reader correctly flagged the VM modeling assumption as weakest, but the load-bearing gap for the quoted strongest claim is the missing downstream experiment. Full-text evaluation sections match the abstract's scope (UQ metrics only), so the integration assertion stays unsupported. This moves the verdict from UNVERDICTED to CONDITIONAL: the UQ comparison may be sound, yet the pipeline benefit requires explicit verification.","tokens_in":1670,"tokens_out":332,"duration_ms":30005,"concrete_test":"Take the reported ENS (mu, kappa) outputs on the paper's test set, plug them into a standard probabilistic data association step (e.g., compute VM likelihoods for gating and weighting), run a simple Kalman-filter tracker, and compare OSPA or MOTA against an EDL-based baseline on the same sequences; a >10% degradation or implementation failure would falsify the facilitation claim.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The strongest claim states that the ENS (mu, kappa) representation 'enables direct probabilistic integration into association modules via closed form VM likelihoods, facilitating a unified detection tracking pipeline.' The manuscript evaluates only upstream DOA uncertainty via risk-coverage curves and AUROC under ID/OOD radar conditions; no association filter (JPDA, etc.), no multi-target tracking metrics, and no end-to-end pipeline results appear. The closed-form property may hold mathematically for the von Mises, yet the 'facilitating' benefit remains an unverified assertion rather than a demonstrated outcome.","agreement_with_reader":"partial"},"referee_report":{"model":"grok-4.3","summary":"The paper compares a von Mises ensemble (ENS) model, which outputs angular DOA predictions parameterized by (μ, κ), against an evidential deep learning (EDL) approach yielding Student-t predictive distributions for uncertainty-aware direction-of-arrival estimation in automotive radar. It reports that ENS achieves lower uncertainty under nominal (in-distribution) conditions and greater sensitivity to severe out-of-distribution perturbations via risk-coverage curves and AUROC analyses, while EDL offers smoother uncertainty variation and slightly better ranking consistency. The manuscript highlights that the ENS (μ, κ) representation permits direct probabilistic integration into association modules through closed-form von Mises likelihoods, thereby facilitating a unified detection-tracking pipeline.","tokens_in":1739,"tokens_out":475,"duration_ms":28836,"significance":"If the empirical comparisons are reproducible and the modeling assumptions hold, the work contributes a geometrically consistent alternative to Euclidean uncertainty quantification for circular DOA problems and surfaces a concrete trade-off between directional fidelity and statistical flexibility. The closed-form VM likelihood property is mathematically attractive for downstream fusion, but its practical benefit for tracking remains an assertion rather than a demonstrated outcome.","major_comments":[{"comment":"Abstract: The claim that 'the ENS representation enables direct probabilistic integration into association modules via closed form VM likelihoods, facilitating a unified detection tracking pipeline' is presented as an important finding, yet the manuscript reports only upstream DOA uncertainty metrics (risk-coverage and AUROC) under ID/OOD conditions; no JPDA-style association filter, multi-target tracking metrics (e.g., MOTA, OSPA), or end-to-end pipeline results are provided to substantiate the facilitation benefit.","section":"Abstract"},{"comment":"Evaluation sections: Performance differences between ENS and EDL are stated without accompanying equations for the ENS loss, the precise definition of the von Mises parameters, dataset sizes, training procedures, or statistical significance tests, making it impossible to verify whether the reported advantages are load-bearing or could be explained by implementation details.","section":"Evaluation"}],"minor_comments":[{"comment":"Notation for the concentration parameter κ and its mapping to uncertainty should be clarified with an explicit equation relating κ to angular variance.","section":null}],"recommendation":"major_revision","confidential_remarks":null},"author_rebuttal":{"model":"grok-4.3","summary":"We thank the referee for the detailed and constructive report. We address each major comment below and will revise the manuscript accordingly to improve clarity and completeness.","responses":[{"response":"We agree that the manuscript provides no end-to-end tracking experiments or association-filter results to empirically substantiate the facilitation claim. The statement rests on the mathematical property that the von Mises (μ, κ) output admits a closed-form likelihood for use in probabilistic association. In revision we will rephrase the abstract to present this as a theoretical advantage of the representation rather than a demonstrated outcome of the current work.","revision_made":"yes","referee_comment":"[Abstract] The claim that 'the ENS representation enables direct probabilistic integration into association modules via closed form VM likelihoods, facilitating a unified detection tracking pipeline' is presented as an important finding, yet the manuscript reports only upstream DOA uncertainty metrics (risk-coverage and AUROC) under ID/OOD conditions; no JPDA-style association filter, multi-target tracking metrics (e.g., MOTA, OSPA), or end-to-end pipeline results are provided to substantiate the facilitation benefit."},{"response":"We acknowledge the need for these details to ensure reproducibility. The revised manuscript will add the explicit ENS loss function, the definition of the von Mises parameters (μ, κ), dataset sizes and splits, training hyperparameters and procedures, and any statistical significance tests performed on the reported metrics.","revision_made":"yes","referee_comment":"[Evaluation] Performance differences between ENS and EDL are stated without accompanying equations for the ENS loss, the precise definition of the von Mises parameters, dataset sizes, training procedures, or statistical significance tests, making it impossible to verify whether the reported advantages are load-bearing or could be explained by implementation details."}],"tokens_in":1374,"tokens_out":388,"duration_ms":26735,"standing_objections":[]},"desk_editor":{"model":"grok-4.3","letter":"The main takeaway is that this work applies an existing circular-statistics ensemble to automotive radar DOA and compares it against an evidential deep-learning baseline on risk-coverage and AUROC metrics under ID and OOD conditions. It shows the von Mises version giving tighter uncertainty on nominal data and sharper response to strong perturbations, while the EDL version is smoother. That is the concrete empirical content.\n\nWhat is new is the side-by-side evaluation on radar angular data with the specific (mu, kappa) parameterization. The geometric alignment of the von Mises model with directional statistics is a reasonable modeling choice for this domain, and the paper correctly notes the closed-form likelihood property. The abstract states the performance differences plainly.\n\nThe soft spot is the downstream claim. The text asserts that the ENS representation enables direct probabilistic integration into association modules and a unified detection-tracking pipeline, yet the reported experiments stop at upstream DOA uncertainty; no JPDA-style filter, no multi-target tracking metrics, and no end-to-end results are shown. That gap makes the integration benefit an assertion rather than a demonstrated outcome. Dataset details, training procedures, and statistical significance tests are also thin in the provided text, which limits how firmly the ranking can be read.\n\nThe work is aimed at radar signal-processing engineers who already work with uncertainty-aware DOA and want a practical comparison of two uncertainty representations. A reader already familiar with both von Mises and evidential methods will find the empirical trade-off useful but will not see a new technique or a closed pipeline.\n\nI would send it to peer review. The comparison is narrow but grounded enough to deserve referee scrutiny on the experimental setup and on whether the integration claim needs to be qualified or supported with additional results.","headline":"The paper runs a comparison of von Mises ensemble versus evidential DL for radar DOA uncertainty but leaves the claimed tracking-pipeline benefit untested.","tokens_in":2232,"tokens_out":421,"would_cite":false,"duration_ms":19636,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"grok-4.3","headline":"Von Mises ensemble for radar DOA produces parameters that integrate directly into tracking via closed-form likelihoods.","keywords":["von Mises distribution","uncertainty quantification","direction of arrival","automotive radar","evidential deep learning","target association","probabilistic tracking"],"falsifier":"A collection of measured angular residuals from real closely spaced targets whose histogram deviates markedly from the von Mises density at the fitted kappa values.","tokens_in":2544,"feed_emoji":"📡","tokens_out":619,"duration_ms":28076,"temperature":0.7,"pith_summary":"The paper compares a circular von Mises ensemble against an evidential deep learning method for producing uncertainty-aware direction-of-arrival estimates in automotive radar. The ensemble outputs a mean direction and concentration parameter that respects angular geometry, while the alternative yields a Student-t distribution in Euclidean space. Under nominal and perturbed conditions the ensemble shows lower nominal uncertainty and sharper response to severe shifts, yet the key advance is that its representation supplies closed-form von Mises likelihoods for immediate use in association steps. This setup supports a single probabilistic pipeline from detection through tracking rather than separate uncertainty handling stages.","feed_headline":"Von Mises parameters plug directly into radar trackers","feed_subtitle":"Ensemble outputs mean angle and concentration that supply closed-form likelihoods for association without extra approximations.","key_machinery":"Von Mises ensemble (ENS) that outputs direction-of-arrival predictions as a mean angle mu and concentration kappa.","core_discovery":"The ENS framework produces angular predictions parameterized by (mu, kappa), enabling interpretable uncertainty aligned with directional geometry. Performance is evaluated under in-distribution and multiple out-of-distribution conditions using risk coverage and ROC or AUROC analyses. Results indicate that ENS achieves lower uncertainty under nominal conditions and exhibits stronger sensitivity to severe perturbations, whereas EDL provides smoother uncertainty variation and slightly improved ranking consistency. Importantly, the ENS representation enables direct probabilistic integration into association modules via closed form VM likelihoods, facilitating a unified detection tracking pipelin","pith_inferences":["The same (mu, kappa) outputs could be fed into multi-hypothesis trackers to maintain angular separation hypotheses without discretizing the circle.","Because likelihoods are analytic, the approach may reduce the need for Monte-Carlo sampling inside downstream fusion filters.","The geometric consistency of the representation suggests it could be paired with other circular statistics modules such as circular filtering for ego-motion compensation."],"forward_implications":["ENS yields lower uncertainty values than EDL under nominal operating conditions.","ENS uncertainty responds more strongly to severe distribution shifts than EDL.","EDL uncertainty varies more smoothly across conditions and ranks detections slightly more consistently.","Closed-form VM likelihoods allow association modules to consume ENS outputs without additional approximation steps."],"fun_headline_variants":["Von Mises ensemble feeds closed-form likelihoods to trackers","Mu and kappa enable radar target association without approximations","ENS provides directional uncertainty for automotive radar DOA","VM parameters integrate into unified radar detection tracking","Von Mises models angular uncertainty in closely spaced targets"],"cache_read_input_tokens":2112,"weakest_assumption_plain":"The von Mises distribution faithfully represents the angular uncertainty that arises when radar targets are closely spaced.","fun_headline_variants_meta":{"raw":{"variants":["Von Mises ensemble feeds closed-form likelihoods to trackers","Mu and kappa enable radar target association without approximations","ENS provides directional uncertainty for automotive radar DOA","VM parameters integrate into unified radar detection tracking","Von Mises models angular uncertainty in closely spaced targets"]},"model":"grok-4.3","cost_usd":0.003922,"raw_usage":{"total_tokens":1995,"prompt_tokens":637,"num_sources_used":0,"completion_tokens":69,"cost_in_usd_ticks":39224500,"prompt_tokens_details":{"text_tokens":637,"audio_tokens":0,"image_tokens":0,"cached_tokens":256},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":1289,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":637,"tokens_out":69,"duration_ms":19384,"temperature":1.0,"reasoning_tokens":1289,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-07-01T03:52:54.135135+00:00","model_set":{"reader":"grok-4.3"},"falsifier":"A collection of measured angular residuals from real closely spaced targets whose histogram deviates markedly from the von Mises density at the fitted kappa values.","supporting_citations":[],"review_version":1}