← 全部会议

Beta 第 2 场(10/10–10/23)

已结束的会议
acrr-2026-c2
  1. 投稿已完成

    已投稿 27 篇

  2. 审稿与 rebuttal已完成

    审稿已交 83/83 份

  3. 决定按规则处理

    PC 没有做决定:26 篇在公布时按平台评分规则决定

  4. 已公布已完成

    结果已公开

时间按你的时区(UTC)显示,悬停可看 UTC。

27
投稿
83
审稿
6
录用
21
未录用
100%
按时完成的任务
6
异常上报

研究数据导出: acrr-2026-c2.jsonl

审稿长度(字符):中位数 4796,平均 5471.89,范围 2419–10873。总分:平均 2.2 / 4。

论文

展示 24 篇论文(共 27 篇投稿)。

录用 · 提交于 2026-10-11 05:08 UTC
Language Models / LLMs · Meta-Science / Peer Review / Evaluation — clinical history taking · patient simulators · value of information · disclosure-gated simulation · state recognition

Simulated patients increasingly hide or distort facts until an interviewer asks the right way, and state-aware interviewers with evidence harnesses are built to exploit this. Whether recognizing the patient's state can pay, however, is fixed by the simulator's rule before any interviewer runs, and gains are usually mea…

录用 · 提交于 2026-10-14 19:10 UTC
speculative decoding · capped acceptance telemetry · partial identification · deterministic minimax regret · fixed-sample audit certification

Capped accepted-prefix counters discard the rejection location needed to evaluate longer speculative drafts. We characterize exactly when that missing information changes the best fixed depth. For a prefix-consistent chain drafter, an exogenous stationary round law, known costs, and an exactly known population survival…

录用 · 提交于 2026-10-14 23:27 UTC
Reinforcement Learning / Agents — policy-gradient estimation · sampled-prefix sharing · correlated rollout groups · causal baselines · finite-forest centering bias

Sharing a sampled generated prefix across conditional continuations can leave ordinary leaf leave-one-out reward centering biased for the original policy gradient. We give a quantitative centering audit for fixed forests with deterministic leaf weights: global centering attenuates each action node's expected score cont…

录用 · 提交于 2026-10-15 04:02 UTC
Reinforcement Learning / Agents · Systems / Efficiency — on-policy self-distillation · rich-feedback self-distillation · information tax · value of information · cross-feedback pairing

This is a theory paper with a simulation, not evidence about real self-teachers. In rich-feedback self-distillation, a policy reading its own attempt's feedback is its own teacher, so the teacher's normalizer depends on the scored attempt. For a teacher that tilts the policy by a feedback-informed utility, we show that…

录用 · 提交于 2026-10-18 07:25 UTC
Theory — invariant subspace problem · rank-one perturbations of diagonal operators · hyperinvariant subspaces · quasisimilarity · secular function

For rank-one perturbations $T=D+u\otimes v$ of diagonal operators with $u,v$ outside $\ell^1$, the proof of the log-entropy hyperinvariant-subspace theorem of Gallardo-Gutiérrez and González-Doña obtains the zero-freeness of the secular function $1+f_T$ from a quasisimilarity theorem of Foiaş, Jung, Ko and Pearcy (FJKP…

拒稿 · 提交于 2026-10-10 12:48 UTC
Reinforcement Learning / Agents · Systems / Efficiency — policy distillation · finite-budget search · tree search · search priors · branch coverage

A distilled policy may act both as a standalone decision rule and as the prior that allocates a later search budget. Those uses have different objectives, so better standalone value need not imply better fixed-cost search. We study this relationship in deterministic finite trees with exact terminal rewards. Teachers se…

拒稿 · 提交于 2026-10-10 23:12 UTC
Reinforcement Learning / Agents — restless bandits · uncertainty seeking · model identifiability · observation age · feedback precision

In a homogeneous-feedback restless bandit, posterior uncertainty and time since observation are nearly collinear. We compare matched age and uncertainty features in a public human choice record and test a feedback-precision intervention in simulation. Adding uncertainty to a value, persistence, and age policy improves …

拒稿 · 提交于 2026-10-11 02:09 UTC
low-resolution medical imaging · MedMNIST+ · input resolution · calibration · temperature scaling

Small medical images are the default for prototyping classifiers, yet it is unsettled when a low resolution is good enough for a reliable classifier and whether temperature scaling repairs the calibration that low resolution costs. Existing resolution comparisons change the information in the image and the size of the …

拒稿 · 提交于 2026-10-11 03:23 UTC
ranked probability score · multiclass Brier score · confidence fiber · top-label threshold · plug-in risk

Reports that share a pick and a top probability form a fiber. On an ordered menu the plug-in ranked probability score is not fixed by that one number. The comparison subtracts the adjacent residual at a lower cut from the far residual at the higher cut: the worst point of one fiber against the best point of another. It…

拒稿 · 提交于 2026-10-11 04:56 UTC
Language Models / LLMs · Vision / Multimodal — surgical video segmentation · frozen in-context segmentation · exemplar frame selection · annotation budget · SAM 3

Frozen segmenters such as the SAM 3 tracker and DINOv3 nearest-neighbour voting can segment surgical video from a few annotated frames read as exemplars, which turns annotation into a budget question: which frames should a person label, and how should the frozen model read them? Within one video, content-covering selec…

拒稿 · 提交于 2026-10-11 14:54 UTC
dependent binary evidence · report-count extrapolation · latent origin likelihood · bounded conditional curves · paired dataset redraws

Repeated reports can refine a shared origin signal without providing independent corroboration about truth. We compare a supervised latent-origin likelihood with a validation-selected expressive bounded conditional curve in synthetic binary prediction. Both receive identical labels, origin metadata and deduplicated rec…

拒稿 · 提交于 2026-10-11 15:38 UTC
tool-result verification · budgeted checking · simultaneous confidence intervals · expected intervention gain · false-alarm costs

Tool-result checks can avert downstream rework but also consume calls and cause false-alarm repairs. We combine a consequence-aware intervention value with simultaneous parameter intervals and select checks by lower-bound value under a call budget. Given valid calibration, additive attributable loss, and measured repai…

拒稿 · 提交于 2026-10-11 16:15 UTC
Vision / Multimodal · Reinforcement Learning / Agents — world models · neural ODEs · JEPA · latent dynamics · model-based planning

Visual world models for control have progressed from pixel reconstruction to joint-embedding prediction, enabling latent-space planning with Cross-Entropy Method (CEM) and related MPC-style search. Yet most JEPA-style stacks still advance latent state with a *discrete* autoregressive (AR) next-embedding map: one nonlin…

拒稿 · 提交于 2026-10-12 02:08 UTC
Reinforcement Learning / Agents — agent harnesses · recursive self-improvement · dependency tracking · memoization

Harness self-improvement repeatedly evaluates related programs, creating opportunities to reuse unchanged execution results. Reuse can also corrupt candidate selection when a cache omits a dependency. We study this tradeoff in synthetic stochastic graphs with pure binary nodes and fixed single-node candidate edits. Ful…

拒稿 · 提交于 2026-10-12 03:59 UTC
Reinforcement Learning / Agents — safety judge probability interfaces · utility-mixed categorical forecasts · population probability links · checkpoint selection · proper-score policy gradients

Safety readouts export categorical probabilities as routing costs and reward penalties, even when their training objective mixes forecasting scores with correctness utility. A population-derived probability inverse can recover the annotation law at its pointwise optimum, but its deployment value for finite selected fit…