NeurIPS 2026 论文列表

Rare disease diagnosis is a fundamental challenge due to heterogeneous and overlapping clinical phenotypes. While recent LLM-based agentic systems have demonstrated promise by integrating external medical tools and knowledge, they typically rely on large-scale or commercial LLMs, limiting their practical deployment in resource-constrained settings. In this work, we observe that their performance significantly degrades when using small-scale LLMs, due to entangled workflows and unreliable multi-evidence aggregation. To address this issue, we propose RADAR, a rare disease diagnostic framework designed for small-scale LLMs. RADAR adopts a divide-and-conquer paradigm that decouples heterogeneous diagnostic evidence into specialized workflows, reducing reasoning complexity and improving robustness. It further introduces a knowledge-driven self-evaluation mechanism that performs evidence-aware reasoning over structured disease knowledge to produce interpretable reliability scores by identifying key supporting phenotypes and conflicting evidence. Finally, a multi-evidence fusion strategy integrates outputs from multiple sources based on reliability and clinical agreement, resolving conflicts and producing stable diagnostic rankings. Experiments on three benchmarks show that RADAR consistently outperforms various state-of-the-art baselines, including general and medical LLMs and agentic systems. Code will be released.

The growing context lengths of large language models (LLMs) have made key-value (KV) cache memory a critical bottleneck during inference. Vector quantization (VQ) with commutative codebooks provides an effective Rotary Position Embedding (RoPE)-aware solution by enabling reconstruction-free attention over quantized key indices. However, existing commutative VQ methods still parameterize scalar codebook coefficients across RoPE subspaces and codewords largely independently, leaving parameter-side redundancy underexploited. We propose $\textbf{RankVQ}$, a low-rank parameterized commutative vector quantization method for KV cache compression. Instead of factorizing the deployed codebooks, RankVQ factorizes the underlying scalar coefficient matrices used to construct key and value codebooks. This design preserves the pseudo-symmetric structure required for RoPE commutativity on the key side, while improving the quality and compactness of offline codebook construction. The low-rank factors are discarded after codebook construction, so RankVQ keeps the same online inference form as standard commutative VQ and introduces no additional decoding overhead. Experiments on LongBench, InfiniteBench, and GSM8K show that RankVQ achieves a stronger accuracy--compression trade-off than directly comparable KV quantization baselines, with particularly clear gains in the 1-bit regime.

Discrete autoregressive (AR) text-to-image (T2I) models adopt a two-stage paradigm in which a VQ tokenizer maps images to discrete codes and an AR policy models their distribution. Current post-training methods optimize only the AR policy while keeping the VQ decoder frozen. We show that this practice introduces Latent Covariate Shift: as the policy evolves, its token distribution progressively diverges from the ground-truth distribution on which the decoder was trained, such that reward scores improve while decoded image quality degrades. To address this mismatch, we propose **RankE**, the first end-to-end post-training framework for discrete T2I generation. Rather than optimizing the policy against a fixed decoder, RankE co-evolves the policy and decoder through alternating optimization: the policy is refined via group-relative preference optimization, while the decoder is jointly adapted via reward-aware adversarial training. This co-evolution ends the fidelity--alignment trade-off that plagues frozen-decoder approaches: on LlamaGen-XL (775M), standard RL improves CLIP but degrades FID, whereas RankE simultaneously improves both (FID 15.21, CLIP 33.76 on MS-COCO 30K). Consistent joint gains on Janus-Pro (1B) confirm that decoder co-evolution reliably converts reward optimization into pixel-space quality improvements.

Deep learning models trained to optimize average accuracy often exhibit systematic failures on particular subpopulations. In real-world settings like healthcare, the subpopulations most affected by such disparities are frequently unlabeled, partially observed, or not known in advance. Existing group-robust methods typically assume prior knowledge of the relevant subgroups, using group annotations for training, validation, or model selection. We propose Misclassification Aware Rank-Limited Adaptation (MARLA), a parameter-efficient method for improving worst group performance without explicit subgroup annotations. MARLA leverages an ERM-trained model by calculating the model's misclassification probability scores on a held-out adaptation set to identify a low-dimensional subspace where errors concentrate. We then learn a rank-restricted additive correction to the classifier logits within that subspace. Across seven real-world datasets, we evaluate group robustness under three settings: no knowledge of subgroup relevance, partial knowledge of subgroup relevance, and full knowledge of subgroup relevance. MARLA improves worst-group performance while remaining fast, parameter-efficient, and practical to tune.

Graph representation learning has become a standard approach for analyzing networked data, with latent embeddings widely used for link prediction, community detection, and related tasks. Yet a basic design choice, the latent dimension, is still treated as a brittle hyperparameter, fixed before training and tuned by held-out performance. Learned factors are also identifiable only up to rotation and rescaling, so the nominal rank rarely coincides with the quantity that governs model behavior. We propose Spectral Prefix Extraction and Capacity-Targeted Representation Analysis (Spectra), which replaces rank as the unit of analysis with the spectrum of a learned positive semidefinite kernel, trace-normalized so that spectra are comparable across fits. The normalized eigenvalues form a distribution on the simplex, and their Shannon effective rank acts both as a summary of learned capacity and as a controllable training-time coordinate: a single scalar shapes this realized dimension during training, and bisection targets any desired value within the rank cap. To theoretically support that, we show local regularity and monotonicity of the realized-dimension profile. Across collaboration, social, biological, and infrastructure networks, Spectra traces performance--capacity frontiers that make the trade-off between predictive accuracy and realized dimension visible. It performs competitively with strong link-prediction baselines, yields aligned lower-capacity views of the same fitted model through spectral prefixes, and provides a principled handle on capacity in the overparameterized regime. Capacity thus becomes a property of the fitted model rather than a hyperparameter of the training.

We consider the optimization problem of an expensive-to-evaluate black-box function, in which we can obtain noisy function values in parallel. For this problem, parallel Bayesian optimization (PBO) is a promising approach, which aims to optimize with fewer function evaluations by selecting a diverse input set for parallel evaluation. However, existing PBO methods suffer from poor practical performance or lack theoretical guarantees. In this study, we propose a PBO method, called randomized kriging believer (KB), based on a well-known KB heuristic and inheriting the advantages of the original KB: low computational complexity, a simple implementation, versatility across various BO methods, and applicability to asynchronous parallelization. Furthermore, we show that our randomized KB achieves Bayesian expected regret guarantees. We demonstrate the effectiveness of the proposed method through experiments, including those on real-data emulators.

Vanilla SVGD is known to have a computational cost of $\mathcal{O}(N^2)$ per iteration. In this work, we introduce Random-Projection Tree Stein Variational Gradient Descent (RP-SVGD) to alleviate this computational challenge. This is achieved by restricting kernel interactions to spatially proximal particles clustered via a random projection tree, further reducing to a cost of $\mathcal{O}(N d(\log (N/C)+C))$, where $C$ is a hyperparameter representing the maximum leaf node capacity of the spanning tree. To establish theoretical validity, we introduce a smoothed RP-Tree kernel for any fixed tree realization and prove that it belongs to the Stein class of the target distribution, thereby ensuring the generation of valid gradient flows. In addition, by considering the effective kernel as the expectation of random tree partitions, we verify that it preserves regularity along with other necessary conditions, which guarantee convergence under the approximate gradient flow framework. Extensive experimental results demonstrate that RP-SVGD tends to have competitive performance and significant speedups across various tasks.

Domain generalization methods train on multiple source environments but usually deploy a single pooled predictor. We study a different deployment object: one head per source environment on a shared representation, combined by fixed weights when the test environment label is unavailable. Under a same-meta random-effects model, where environment-specific population minimizers are iid deviations around a shared meta-mean, the optimal fixed simplex rule is the uniform centroid. The same analysis separates this centroid from pooled ERM: under a common feature second moment, pooled ERM is sample-weighted, while the centroid is environment-weighted, giving a closed-form gap controlled by sample-weight imbalance. We introduce **REC** (Random-Effects Centroid), a training objective that fits source-specific experts while directly scoring their averaged deployment head. In the common-second-moment case, the centroid term controls the trained-centroid discrepancy and the dispersion of environment-specific optima. Synthetic experiments isolate the population estimator effect and learned-representation behavior; WILDS experiments show that REC is competitive with established DG baselines and that the centroid term is necessary for a useful averaged predictor.

Neural networks with randomly generated hidden weights (RaNNs) have been extensively studied, both as a standalone learning method and as an initialization for fully trainable deep learning methods. In this work, we study RaNN expressivity for learning solutions to non-linear partial differential equations (PDEs). Despite their widespread use in practical applications, a rigorous theoretical understanding of the approximation properties of RaNNs in this context remains limited. Here, we derive error bounds for RaNN approximations to time-dependent Sobolev functions and obtain a dimension-free approximation rate $\frac{1}{2}$ for sufficiently regular functions. We apply our results to two important classes of non-linear PDEs: Porous Medium Equations and Compressible Navier-Stokes Equations, showing that RaNNs are capable of efficiently approximating solutions to these complex, non-linear PDEs. Our theoretical analysis is supported by numerical experiments, validating the obtained convergence rates.

Empirical studies of trained models often report a transient regime in which signal is detectable in a finite gradient descent time window before overfitting dominates. We provide an analytically tractable random-matrix model that reproduces this phenomenon for gradient flow in a linear teacher--student setting. In this framework, learning occurs when an isolated eigenvalue separates from a noisy bulk, before eventually disappearing in the overfitting regime. The key ingredient is anisotropy in the input covariance, which induces fast and slow directions in the learning dynamics. In a two-block covariance model, we derive the full time-dependent bulk spectrum of the symmetrized weight matrix through a $2\times 2$ Dyson equation, and we obtain an explicit outlier condition for a rank-one teacher via a rank-two determinant formula. This yields a transient Baik--Ben Arous--P\'ech\'e (BBP) transition: depending on signal strength and covariance anisotropy, the teacher spike may never emerge, emerge and persist, or emerge only during an intermediate time interval before being reabsorbed into the bulk. We map the corresponding phase diagrams and validate the theory against finite-size simulations. Our results provide a minimal solvable mechanism for early stopping as a transient spectral effect driven by anisotropy and noise.

Neural scaling laws for transformer language models predict smooth improvements in pretraining loss with increasing parameters, but downstream capabilities such as in-context learning are known to emerge abruptly past a certain model scale. In this paper, we show that emergent capabilities arise at variable intervals throughout training, with larger models acquiring capabilities earlier on average. We demonstrate that the emergence of capabilities such as pattern completion and indirect object identification corresponds to the abrupt learning of task-relevant attention patterns. To isolate this phenomenon, we train transformer models on synthetic linear map and cellular automata datasets, and we show that the difficulty of learning attention patterns depends on context length and pattern sparsity. Moreover, scaling the number of attention heads improves learning efficiency on our synthetic tasks, while increasing the head dimension yields diminishing returns past a minimum capacity. We additionally investigate architectures with alternative attention mechanisms, showing that MLP-Mixer outperforms a transformer on linear map tasks with complex attention patterns. Our findings provide a mechanistic insight into emergence, showing that downstream capabilities arise abruptly due to the intrinsic difficulty of learning sparse attention patterns in transformer models.

Machine learning (ML) has transformed many scientific fields, yet key applications still lack standardized benchmarks. Raman spectroscopy, a widely used technique for non-invasive molecular analysis, is one such field where progress is limited by fragmented datasets, inconsistent evaluation, and models that fail to capture the structure of spectral data. We introduce RamanBench, the first large-scale, fully reproducible benchmark for ML on Raman spectroscopy, consisting of streamlined data access (https://pypi.org/project/raman-data/), evaluation protocols and code (https://pypi.org/project/raman-bench/), as well as a live leaderboard (https://huggingface.co/spaces/HTW-KI-Werkstatt/RamanBench). It unifies 77 datasets (including 17 first released with this benchmark) across four domains, comprising 325,862 spectra and spanning classification and regression tasks under diverse experimental conditions. We benchmark 28 models under a standardized protocol, including classical methods (e.g., PLS), Raman-specific (e.g., RamanNet), tabular foundation models (TFMs; e.g., TabPFN), and time-series approaches (e.g., ROCKET). TFMs consistently outperform domain-specific and gradient boosting baselines, while time-series models remain competitive. However, no single method dominates across these heterogeneous datasets, revealing that Raman spectroscopy remains an open challenge for machine learning. Therefore, we invite the community to contribute new approaches to our living benchmark, with the potential to accelerate advances in critical applications such as medical diagnostics, biological research, and materials science.

Noisy-label methods often estimate sample reliability from forward-space signals such as loss, confidence, or entropy. These signals indicate whether a sample is difficult to predict, but they do not directly test whether its observed label induces a reliable parameter update. This gap matters because hard clean samples and mislabeled samples can have similar loss while inducing different updates. We recast reliability estimation as diagnosis of the observed-label update. The sample-wise empirical Fisher trace gives a backward-space measure of update energy: for the classifier layer, it factorizes into a prediction-residual term and a feature-sensitivity term, so it captures information beyond scalar loss. Trace, however, is still a radial magnitude signal and cannot decide whether a large update is useful or harmful. We therefore propose Relative Geometric Conflict (RGC), which compares the observed-label gradient with a reference gradient induced by an EMA teacher. The conflict term helps distinguish large but aligned hard-clean updates from large conflicting updates caused by corrupted labels. Across synthetic and real-world noisy-label benchmarks, RGC improves hard-clean preservation and accuracy under our evaluation protocol.

Masked Autoencoders (MAE) have emerged as a powerful self-supervised paradigm in vision, yet their direct application to radar perception remains bottlenecked by a fundamental domain gap. Specifically, while conventional MAEs are optimized for RGB images using isotropic grid patching and standard mean squared error (MSE) loss, radar range-azimuth (RA) heatmaps exhibit sinc-oriented anisotropic spatial structures as well as complex physical artifacts, such as heavy-tailed multipath interference and sidelobe spikes. In this paper, we propose RadarMAE, the first radar-centric MAE tailored specifically to the physical properties of radar signals. We present an axial tokenization strategy that preserves the resolution disparities and continuous sinc-shaped lobes of RA maps. Furthermore, considering that the range and azimuth dimensions exhibit distinct physical artifacts, we decouple their reconstruction objectives by introducing a multipath distribution loss along the range axis to capture heavy-tailed multipath reflections as well as a sidelobe distribution loss along the azimuth axis to learn deterministic sidelobe spikes. Our extensive experiments demonstrate that injecting these domain-specific physical priors enables robust representation learning for radar object detection. Notably, our method achieves a state-of-the-art 64.17% bird's-eye-view (BEV) AP on the K-Radar dataset using only a single-frame RA map setting, significantly outperforming the previous baselines.

Radar-based human pose estimation is a promising alternative to camera-based perception, but remains challenging due to sparse observations, noisy measurements, and severe pose ambiguity. Existing methods infer both global skeletal structure and flexible peripheral joints from sparse radar observations, often compromising structural plausibility and fine-grained local accuracy. Motivated by the coarse-to-fine nature of human visual perception, we propose RadarFlowPose, a framework that decomposes radar-based human pose estimation into coarse pose prior generation and flow-based refinement. The coarse model first produces a structurally plausible and temporally coherent pose prior from sparse radar observations, providing a global initialization for subsequent refinement. Instead of estimating the full pose from scratch, the refinement stage focuses on correcting the remaining local ambiguities and motion-induced errors. By incorporating localised Doppler motion cues into the flow-matching process, the model performs motion-aware residual refinement around the coarse prior, leading to more accurate and anatomically consistent pose estimates. Experiments on three benchmark radar pose datasets show consistent effectiveness, especially in motion-sensitive and fine-grained joint-level evaluation.

Non-contrast CT (NCCT) is widely used in clinical practice, yet its limited soft-tissue contrast often renders visual information insufficient for accurate diagnosis. In routine radiology workflows, NCCT is commonly accompanied by contrast-enhanced CT, radiology reports, and pathology reports, which provide complementary knowledge at different levels, from structural and morpho-functional cues to semantic findings and pathological evidence. However, existing pretraining paradigms address this challenge only partially, each introducing complementary but still limited constraints on NCCT representations. To address this limitation, we propose RadOmni, a unified omni-pretraining framework that injects multimodal radiology knowledge into representation learning for non-contrast CT. RadOmni formulates a curriculum consolidation learning strategy, in which knowledge is progressively injected according to optimization difficulty and knowledge level through four stages: masked image modeling on non-contrast CT, cross-phase transfer of anatomical structures and enhanced patterns from contrast-enhanced CT, report-guided semantic learning, and pathology-informed diagnostic enhancement. To mitigate optimization conflicts across heterogeneous objectives and reduce knowledge forgetting, each stage retains and jointly optimizes the objectives from preceding stages, while early encoder layers are selectively frozen to preserve generic representations. To enable such multi-source pretraining, we curate Omni-CT10K, a large-scale CT dataset with more than 10K NCCT scans featuring clinically co-occurring multimodal supervision. Extensive experiments on Omni-CT10K, two in-house benchmarks from different centers, and the public MSD benchmark show that RadOmni consistently achieves strong performance across classification, segmentation, pathological TNM staging, and radiology report generation tasks. Further analyses demonstrate favorable scaling behavior, robust cross-center generalization, and transferable representations.

The recently introduced 4-bit floating-point format, NVFP4, demonstrates remarkable performance and memory benefits for quantized large language model (LLM) inference. However, we observe two types of redundancy in the existing NVFP4 encoding: (1) The FP4 format naturally exposes an unused quantization value due to its sign-magnitude representation that contains both positive and negative zeros. (2) The FP8 block scaling factor, with 4-bit exponent and 3-bit mantissa, contains an unused sign bit given that the scaling factor is always positive. Additionally, we find that LLM weights are more tolerant to a lower-precision block scaling factor, such as 6 bits with 3-bit exponent and 3-bit mantissa. Based on these observations, we propose **R**edund**a**nt **Ze**ro **R**emapping (RaZeR), an enhanced numerical format that pushes the limits of NVFP4 for more accurate LLM quantization under the same memory footprint. RaZeR leverages the unused bits in the block scaling factor to adaptively remap the negative FP4 zero to a set of pre-defined special values, which maximally utilizes the NVFP4 encoding and better fits the LLM tensor distribution. Extensive experiments validate RaZeR’s superior performance for 4-bit LLM quantization. For example, RaZeR reduces the average perplexity loss of NVFP4 by 29.4% and 32.3% under weight-only and weight-activation quantization, respectively.

Comprehensively retrieving diverse documents is crucial to address queries that admit a wide range of valid answers. We introduce retrieve-verify-retrieve (RVR), a multi-round retrieval framework designed to maximize answer coverage. Initially, a retriever takes the original query and returns a candidate document set, followed by a verifier that identifies a high-quality subset. For subsequent rounds, the query is augmented with previously verified documents to uncover answers that are not yet covered in previous rounds. RVR is effective even with off-the-shelf retrievers, and fine-tuning retrievers for our inference procedure brings further gains. Our method outperforms baselines, including agentic search approaches, achieving at least 10\% relative and 3\% absolute gain in complete recall percentage on a multi-answer retrieval dataset (QAMPARI). We also see consistent gains on two out-of-domain datasets (QUEST and WebQuestionsSP) across different base retrievers. Our work presents a promising iterative approach for comprehensive answer recall leveraging a verifier and adapting retrievers to a new inference scenario.

Modern Voice Cloning (VCL) can synthesize speech that closely matches a target speaker from only seconds of reference audio, enabling applications such as personalized speech interfaces and dubbing. In practical deployments, modern audio generation models inevitably encounter noisy reference audios, imperfect text prompts, multilingual and long-form generation settings, downstream post-processing, and adversarial perturbations, all of which can significantly hurt robustness. Despite rapid progress in VCL driven by autoregressive codec-token language models and diffusion-based models, robustness under realistic deployment shifts remains underexplored. This paper introduces RVCBench, a comprehensive dataset and benchmark that evaluates Robustness in Voice Clone across the full generation pipeline. RVCBench contributes a large-scale, task-aligned robustness dataset that instantiates realistic deployment shifts through controlled text-audio pairing, multilingual and long-form scenarios, expressive prompts, post-processing conditions, and passive or proactive audio perturbations. Covering 18 robustness evaluations, 225 speakers, and 14,370 utterances, RVCBench enables unified evaluation of input sensitivity, generation stability, output resilience, and perturbation robustness. We evaluate 18 representative modern VCL models and reveal systematic vulnerabilities in content consistency, speaker similarity, long-form stability, post-processing resilience, detectability, and adversarial robustness.