ICLR 2026 论文列表

Feedforward models for novel view synthesis (NVS) have recently advanced by transformer-based methods like LVSM, using attention among all input and target views. In this work, we argue that its full self-attention design is suboptimal, suffering from quadratic complexity with respect to the number of input views and rigid parameter sharing among heterogeneous tokens. We propose Efficient-LVSM, a dual-stream architecture that avoids these issues with a decoupled co-refinement mechanism. It applies intra-view self-attention for input views and self-then-cross attention for target views, eliminating unnecessary computation. Efficient-LVSM achieves 29.86 dB PSNR on RealEstate10K with 2 input views, surpassing LVSM by 0.2 dB, with 2× faster training convergence and 4.2× faster inference speed. Efficient-LVSM achieves state-of-the-art performance on multiple benchmarks, exhibits strong zero-shot generalization to unseen view counts, and enables incremental inference with KV-cache, thanks to its decoupled designs.

Query-based image retrieval (QBIR) requires retrieving relevant images given diverse and often stylistically heterogeneous queries, such as sketches, artworks, or low-resolution previews. While large-scale vision--language representation models (VLRMs) like CLIP offer strong zero-shot retrieval performance, they struggle with distribution shifts caused by unseen query styles. In this paper, we propose the Hypernetwork-driven Style-adaptive Retrieval (Hystar), a lightweight framework that dynamically adapts model weights to each query’s style. Hystar employs a hypernetwork to generate singular-value perturbations ($\Delta S$) for attention layers, enabling flexible per-input adaptation, while static singular-value offsets on MLP layers ensure cross-style stability. To better handle semantic confusions across styles, we design StyleNCE as part of Hystar, an optimal-transport-weighted contrastive loss that emphasizes hard cross-style negatives. Extensive experiments on multi-style retrieval and cross-style classification benchmarks demonstrate that Hystar consistently outperforms strong baselines, achieving state-of-the-art performance while being parameter-efficient and stable across styles.

Real-world visual signals are inherently variable in resolution, and it is natural to endow multimodal large language models (MLLMs) with such native-resolution perception capabilities. In principle, for general and straightforward multimodal understanding, low-resolution images are sufficient. While for images with nuanced details like documents and charts, it is crucial to preserve fine-grained details using high-resolution inputs, as naive resizing inevitably results in information loss. Recent advances employ sequence packing to process images of any resolution and aspect ratios. Despite these efforts, model performance degrades at both low and high resolutions, and high-resolution inputs incur substantial computational costs. We argue that the rigid use of a single patch size is the primary cause: when image resolution or information density varies, fixing patch size is intrinsically suboptimal. To address this issue, we introduce Adaptive Patching (AdaPatch), a simple yet effective strategy that adjusts patch size according to image resolution and information density and could be seamlessly plugged into pre-trained fixed-patch MLLMs without any training efforts. Extensive evaluations demonstrate consistent improvements in native resolution performance without additional training. Besides, we provide a training-based method to further adapt MLLMs with dynamic patch sizes and enhance the performance.

Unified Speech Recognition (USR) has emerged as a semi-supervised framework for training a single model for audio, visual, and audiovisual speech recognition, achieving state-of-the-art results on in-distribution benchmarks. However, its reliance on autoregressive pseudo-labelling makes training expensive, while its decoupled supervision of CTC and attention branches increases susceptibility to self-reinforcing errors, particularly under distribution shifts involving longer sequences, noise, or unseen domains. We propose CTC-driven teacher forcing, where greedily decoded CTC pseudo-labels are fed into the decoder to generate attention targets in a single forward pass. Although these can be globally incoherent, in the pseudo-labelling setting they enable efficient and effective knowledge transfer. Because CTC and CTC-driven attention pseudo-labels have the same length, the decoder can predict both simultaneously, benefiting from the robustness of CTC and the expressiveness of attention without costly beam search. We further propose mixed sampling to mitigate the exposure bias of the decoder relying solely on CTC inputs. The resulting method, USR 2.0, halves training time, improves robustness to out-of-distribution inputs, and achieves state-of-the-art results on LRS3, LRS2, and WildVSR, surpassing USR and modality-specific self-supervised baselines.

Conditional Semantic Textual Similarity (C-STS) measures the semantic proximity between text segments under a specific condition, thereby overcoming the ambiguity inherent in traditional STS. However, existing methods are largely confined to discriminative models, failing to fully integrate recent breakthroughs in the NLP community concerning Large Language Models (LLMs) and Reinforcement Learning (RL). RL is a particularly well-suited paradigm for this task, as it can directly optimize the non-differentiable Spearman ranking metric and guide the reasoning process required by C-STS. However, we find that naively applying listwise RL fails to produce meaningful improvements, as the model is overwhelmed by a complex, coarse-grained reward signal. To address this challenge, we introduce PoLi-RL, a novel Point-to-List Reinforcement Learning framework. PoLi-RL employs a two-stage curriculum: it first trains the model with simple pointwise rewards to establish fundamental scoring capabilities, then transitions to a hybrid reward that combines pointwise, pairwise, and listwise objectives to refine the model's ability to discern subtle semantic distinctions. Crucially, we propose an innovative Parallel Slice Ranking Reward (PSRR) mechanism that computes ranking rewards in parallel slices, where each slice comprises same-indexed completions from different samples. This provides a precise, differentiated learning signal for each individual completion, enabling granular credit assignment and effective optimization. On the official C-STS benchmark, PoLi-RL achieves a Spearman correlation coefficient of 48.18, establishing a new SOTA for the cross-encoder architecture. As the first work to successfully apply RL to C-STS, our study introduces a powerful and effective paradigm for training LLMs on complex, ranking-based conditional judgment tasks. Our code and checkpoints are available at https://anonymous.4open.science/r/PoLi-RL.

Large Audio Language Models (LALMs) represent an important frontier in multimodal AI, addressing diverse audio tasks. Recently, post-training of LALMs has received increasing attention due to significant performance improvements over foundation models. While single-stage post-training such as reinforcement learning (RL) has demonstrated promising results, multi-stage approaches such as supervised fine-tuning (SFT) followed by RL remain suboptimal. The allocation of data across multiple training stages to maximize LALM capabilities has not been fully explored, and large-scale, high-quality datasets for such research are also lacking. To address these problems, we firstly present AudioMCQ, a comprehensive audio multiple-choice question dataset comprising 571k samples with two kinds of chain-of-thought annotations. Secondly, we investigate the prevalent zero audio-contribution phenomenon in LALMs, where models derive correct answers solely from textual information without processing audio content. We propose Audio-Contribution Filtering to partition data into weak and strong audio-contribution subsets. Based on these insights, we develop two effective post-training paradigms: Weak-to-Strong (SFT on weak audio-contribution data followed by RL on strong audio-contribution data) and Mixed-to-Strong (SFT on mixed audio-contribution data followed by RL on strong audio-contribution data). We achieve first place in the DCASE 2025 Audio-Question-Answering challenge by using AudioMCQ. Additionally, leveraging our dataset with different training strategies, we achieve 78.2\% on MMAU-test-mini, 75.6\% on MMAU, 67.0\% on MMAR, and 71.7\% on MMSU, establishing new state-of-the-art performance.

Parameterized and coupled partial differential equations (PDEs) are central to modeling phenomena in science and engineering, yet neural operator methods that address both aspects remain limited. We extend Fourier neural operators (FNOs) with minimal architectural modifications along two directions. For parameterized dynamics, we propose a hypernetwork-based modulation that conditions the operator on physical parameters. For coupled systems, we conduct a systematic exploration of architectural choices, examining how operator components can be adapted to balance shared structure with cross-variable interactions while retaining the efficiency of standard FNOs. Evaluations on benchmark PDEs, including the one-dimensional capacitively coupled plasma equations and the Gray–Scott system, show that our methods achieve up to 55~72% lower errors than strong baselines, demonstrating the effectiveness of principled modulation and systematic design exploration.

Visually localizing an image, i.e., estimating its camera pose, requires building a scene representation that serves as a visual map. The representation we choose has direct consequences towards the practicability of our system. Even when starting from mapping images with known camera poses, state-of-the-art approaches still require hours of mapping time in the worst case, and several minutes in the best. This work raises the question whether we can achieve competitive accuracy much faster. We introduce FastForward, a method that creates a map representation and relocalizes a query image on-the-fly in a single feed-forward pass. At the core, we represent multiple mapping images as a collection of features anchored in 3D space. FastForward utilizes these mapping features to predict image-to-scene correspondences for the query image, enabling the estimation of its camera pose. We couple FastForward with image retrieval and achieve state-of-the-art accuracy when compared to other approaches with minimal map preparation time. Furthermore, FastForward demonstrates robust generalization to unseen domains, including challenging large-scale outdoor environments.

The ability to estimate the quality of scientific papers is central to how both humans and AI systems will advance scientific knowledge in the future. However, existing LLM-based estimation methods suffer from high inference cost, whereas the faster direct score regression approach is limited by scale inconsistencies. We present NAIPv2, a debiased and efficient framework for paper quality estimation. NAIPv2 employs pairwise learning within domain-year groups to reduce inconsistencies in reviewer ratings and introduces the Review Tendency Signal (RTS) as a probabilistic integration of reviewer scores and confidences. To support training and evaluation, we further construct NAIDv2, a large-scale dataset of 24,276 ICLR submissions enriched with metadata and detailed structured content. Trained on pairwise comparisons but enabling efficient pointwise prediction at deployment, NAIPv2 achieves state-of-the-art performance (78.2\% AUC, 0.432 Spearman), while maintaining scalable, linear-time efficiency at inference. Notably, on unseen NeurIPS submissions, it further demonstrates strong generalization, with predicted scores increasing consistently across decision categories from Rejected to Oral. These findings establish NAIPv2 as a debiased and scalable framework for automated paper quality estimation, marking a step toward future scientific intelligence systems.

End-to-end human animation with rich multi-modal conditions, e.g., text, image and audio has achieved remarkable advancements in recent years. However, most existing methods could only animate a single subject and inject conditions in a global manner, ignoring scenarios that multiple concepts could appears in the same video with rich human-human interactions and human-object interactions. Such global assumption prevents precise and per-identity control of multiple concepts including humans and objects, therefore hinders applications. In this work, we discard the single-entity assumption and introduce a novel framework that enforces strong, region‑specific binding of conditions from modalities to each identity's spatiotemporal footprint. Given reference images of multiple concepts, our method could automatically infer layout information by leveraging a mask predictor to match appearance cues between the denoised video and each reference appearance. Furthermore, we inject local audio condition into its corresponding region to ensure layout-aligned modality matching in a iterative manner. This design enables the high-quality generation of human dialogue videos between two to three people or video customization from multiple reference images. Empirical results and ablation studies validate the effectiveness of our explicit layout control for multi-modal conditions compared to implicit counterparts and other existing methods.

The advent of Vision-Language Models (VLMs) has significantly advanced end-to-end autonomous driving, demonstrating powerful reasoning abilities for high-level behavior planning tasks. However, existing methods are often constrained by a passive perception paradigm, relying solely on text-based reasoning. This passivity restricts the model’s capacity to actively seek crucial visual evidence when faced with uncertainty. To address this, we introduce DriveAgent-R1, an autonomous driving agent capable of active perception for planning. In complex scenarios, DriveAgent-R1 proactively invokes tools to perform visual reasoning, firmly grounding its decisions in visual evidence, thereby enhancing both interpretability and reliability. Furthermore, we propose a hybrid thinking framework, inspired by human driver cognitive patterns, allowing the agent to adaptively switch between efficient text-only reasoning and robust tool-augmented visual reasoning based on scene complexity. This capability is cultivated through a three-stage progressive training strategy, featuring a core Cascaded Reinforcement Learning (Cascaded RL) phase. Extensive experiments on the Drive-Internal dataset, which is rich in long-tail scenarios, and the public nuScenes dataset show that, with only 3B parameters, DriveAgent-R1 achieves competitive performance comparable to top closed model systems such as GPT-5 and to human driving proficiency while remaining deployment-friendly, offering a proven path toward building more intelligent autonomous driving systems.

Multimidal Large Language Models (MLLMs) have demonstrated impressive capabilities in textual and 2D visual reasoning, yet their ability to understand and reason over 3D data remains limited. The issues become more challenging for understanding standalone 3D point cloud due to the high interclass confusion. In this work, we propose Point-Graph LLM (PGLLM), a framework that enables more effective 3D point cloud understanding by integrating in-context prompting and score refinement at test-time, respecting supporting data manifold. Our method first employs a pre-trained point cloud encoder which are used to construct a graph where edges encode visual similarity. Each support point cloud sample is converted to a textual caption via pre-trained PointLLM. For a test query, the graph is used to retrieve relevant neighbors whose captions serve as contextual demonstrations for a second stage LLM for final reasoning, a process we term in-context guidance. Furthermore, we introduce a confidence score refinement mechanism based on label propagation to enhance the reliability of LLM predictions for classification and out-of-distribution (OOD) detection tasks. All above optimizations are carried out fully at test-time. Extensive experiments across diverse 3D datasets and tasks demonstrate that PGLLM consistently improves accuracy and robustness over prior baselines with very almost no additional computation cost, showcasing a promising direction toward native 3D reasoning with MLLMs.

Audio and music generation based on flexible multimodal control signals is a widely applicable topic, with the following key challenges: 1) a unified multimodal modeling framework, and 2) large-scale, high-quality training data. As such, we propose AudioX, a unified framework for anything-to-audio generation that integrates varied multimodal conditions (i.e., text, video, image, and audio signals) in this work. The core design in this framework is a Multimodal Adaptive Fusion module, which enables the effective fusion of diverse multimodal inputs, enhancing cross-modal alignment and improving overall generation quality. To train this unified model, we construct a large-scale, high-quality dataset, IF-caps, comprising over 7 million samples curated through a structured data annotation pipeline. This dataset provides comprehensive supervision for multimodal-conditioned audio generation. We benchmark AudioX against state-of-the-art methods across a wide range of tasks, finding that our model achieves superior performance, especially in text-to-audio and text-to-music generation. These results demonstrate our method is capable of audio generation under multimodal control signals, showing powerful instruction-following potential. We will release the code, model, and dataset.

We introduce TangoFlux, an efficient Text-to-Audio (TTA) generative model with 515M parameters, capable of generating up to 30 seconds of 44.1kHz audio in 3.7 seconds on a A40 GPU. A key challenge in aligning TTA models lies in creating preference pairs, as TTA lacks structured mechanisms like verifiable rewards or gold-standard answers available for Large Language Models (LLMs). To address this, we propose CLAP-Ranked Preference Optimization (CRPO), a novel framework that iteratively generates and optimizes preference data to enhance TTA alignment. We show that the audio preference dataset generated using CRPO outperforms the static alternatives. With this framework, TangoFlux achieves state-of-the-art performance across both objective and subjective benchmarks. https://tangoflux.github.io/ holds the model-generated audio samples for comparison.

We propose GOLDILOCS: a novel zero-shot, pose-agnostic method for object-level semantic change detection in the wild. While supervised Scene Change Detection (SCD) methods achieve impressive results on curated datasets, these models do not generalize and performance drops on out-of-domain data. Recent Zero-Shot SCD methods introduced a more robust approach with foundational models as backbone, yet they neglect the 3D aspect of the task and remain constrained to the image-pair setting. Conversely, 3D-centric SCD methods based on 3D Gaussian Splatting (3DGS) or NeRFs require multi-view inputs, but cannot operate on an image pair. Our key insight is that SCD can be reformulated as a 3D reconstruction problem over time, where geometric inconsistencies naturally indicate change. Although previous work considered viewpoint difference a challenge, we recognize the additional geometric information as an advantage. GOLDILOCS uses dense stereo reconstruction to estimate camera parameters and generate a pointmap of the commonalities between input images by filtering geometric inconsistencies. Rendering the canonical scene representation from multiple viewpoints yields reference images that exclude changed or occluded content. Rigid object changes are then detected through mask tracking, while nonrigid transformations are identified using SSIM heatmaps. We evaluate our method on a variety of datasets, covering both pairwise and multi-view cases in binary and multi-class settings, and demonstrate superior performance over prior work, including supervised methods.

Inspired by the brain's spike-based computation, spiking neural networks (SNNs) inherently possess temporal activation sparsity. However, when it comes to the sparse training of SNNs in the structural connection domain, existing methods fail to achieve ultra-sparse network structures without significant performance loss, thereby hindering progress in energy-efficient neuromorphic computing. This limitation presents a critical challenge: how to achieve high levels of structural connection sparsity while maintaining performance comparable to fully connected networks. To address this challenge, we propose the Cannistraci-Hebb Spiking Neural Network (CH-SNN), a novel and generalizable dynamic sparse training framework for SNNs consisting of four stages. First, we propose a sparse spike correlated topological initialization (SSCTI) method to initialize a sparse network based on node correlations. Second, temporal activation sparsity and structural connection sparsity are integrated via a proposed sparse spike weight initialization (SSWI) method. Third, a hybrid link removal score (LRS) is applied to prune redundant weights and inactive neurons, improving information flow. Finally, the CH3-L3 network automaton framework inspired by Cannistraci-Hebb learning theory is incorporated to perform link prediction for potential synaptic regrowth. These mechanisms enable CH-SNN to achieve sparsification across all linear layers. We have conducted extensive experiments on six datasets including CIFAR-10 and CIFAR-100, evaluating various network architectures such as spiking convolutional neural networks and Spikformer. The proposed method achieves a maximum sparsity of 97.75% and outperforms the fully connected (FC) network by 0.16% in accuracy. Furthermore, we apply CH-SNN within an SNN training algorithm deployed on an edge neuromorphic processor. The experimental results demonstrate that, compared to the FC baseline without CH-SNN, the sparse CH-SNN architecture achieves up to 98.84% sparsity, an accuracy improvement of 2.27%, and a 97.5$\times$ reduction in synaptic operations, and the energy consumption is reduced by an average of 55$\times$ across four datasets. To comply with double-blind review requirements, our code will be made publicly available upon acceptance.

Convolutional Neural Networks (CNNs) rely on fixed-size kernels scanning local patches, which limits their ability to capture global context or long-range dependencies without very deep architectures. Vision Transformers (ViTs), in turn, provide global connectivity but lack spatial inductive bias, depend on explicit positional encodings, and remain tied to the initial patch size. Bridging these limitations requires a representation that is both structured and global. We introduce SONIC (Spectral Oriented Neural Invariant Convolutions), a continuous spectral parameterisation that models convolutional operators using a small set of shared, orientation-selective components. These components define smooth responses across the full frequency domain, yielding global receptive fields and filters that adapt naturally across resolutions. Across synthetic benchmarks, large-scale image classification, and 3D medical datasets, SONIC shows improved robustness to geometric transformations, noise, and resolution shifts, and matches or exceeds convolutional, attention-based, and prior spectral architectures with an order of magnitude fewer parameters. These results demonstrate that continuous, orientation-aware spectral parameterisations provide a principled and scalable alternative to conventional spatial and spectral operators.

Multi-Resolution Hash Encoding (MHE), the foundational technique behind Instant Neural Graphics Primitives, provides a powerful parameterization for neural fields. However, its spatial behavior lacks rigorous understanding from a physical systems perspective, leading to reliance on heuristics for hyperparameter selection. This work introduces a novel analytical approach that characterizes MHE by examining its Point Spread Function (PSF), which is analogous to the Green's function of the system. This methodology enables a quantification of the encoding's spatial resolution and fidelity. We derive a closed-form approximation for the collision-free PSF, uncovering inherent grid-induced anisotropy and a logarithmic spatial profile. We establish that the idealized spatial bandwidth, specifically the Full Width at Half Maximum (FWHM), is determined by the average resolution, $N_{\text{avg}}$. This leads to a counterintuitive finding: the effective resolution of the model is governed by the broadened empirical FWHM (and therefore $N_{\text{avg}}$), rather than the finest resolution $N_{\max}$, a broadening effect we demonstrate arises from optimization dynamics. Furthermore, we analyze the impact of finite hash capacity, demonstrating how collisions introduce speckle noise and degrade the Signal-to-Noise Ratio (SNR). Leveraging these theoretical insights, we propose Rotated MHE (R-MHE), an architecture that applies distinct rotations to the input coordinates at each resolution level. R-MHE mitigates anisotropy while maintaining the efficiency and parameter count of the original MHE. This study establishes a methodology based on physical principles that moves beyond heuristics to characterize and optimize MHE.

Understanding human behavior requires measuring behavioral actions. Due to its complexity, behavior is best mapped onto a rich, semantic structure such as language. Emerging multimodal large language models (MLLMs) are promising candidates, but their fine-grained action understanding ability has not been fully examined. In this work, we reformulate EPIC-KITCHENS-100, one of the largest and most challenging egocentric action recognition datasets, into a MLLM benchmark (EPIC-KITCHENS-100-MQA). We show that when we sample difficult answers based on specialist models as distractors, leading MLLMs struggle to recognize the correct actions. How can we increase the performance of MLLMs? We curated a supervised finetuning dataset that includes `hard' action recognition, temporal detection, captioning, and free-form question answering to improve models' diverse action understanding capabilities. We introduce a new model called LLaVAction that adds an action token to boost models' attention on visual tokens and a two-stage pipeline to obtain structured actions. LLaVAction greatly improves the MLLMs' ability of action understanding, achieving strong improvements on both MLLM benchmarks (21 points in accuracy over GPT-4o on EPIC-KITCHENS-100-MQA) and action recognition benchmarks, suggesting that our methods prepare MLLMs to be a promising path forward for complex action tasks. Code, data, benchmark and models will be available upon acceptance.