AAAI 2025 Papers

Following formatting instructions to generate well-structured content is a fundamental yet often unmet capability for large language models (LLMs). To study this capability, which we refer to as format faithfulness, we present FormatBench, a comprehensive format-related benchmark. Compared to previous format-related benchmarks, FormatBench involves a greater variety of tasks in terms of application scenes (traditional NLP tasks, creative works, autonomous agency tasks), human-LLM interaction styles (single-turn instruction, multi-turn chat), and format types (inclusion, wrapping, length, coding). Moreover, each task in FormatBench is attached with a format checker program. Extensive experiments on the benchmark reveal that state-of-the-art open- and closed-source LLMs still suffer from severe deficiency in format faithfulness. By virtue of the decidable nature of formats, we propose to Reinforce Format Faithfulness (ReFF) to help LLMs generate formatted output as instructed without compromising general quality. Without any annotated data, ReFF can substantially improve the format faithfulness rate (e.g., from 21.6% in original LLaMA3 to 95.0% on caption segmentation task), while keep the general quality comparable (e.g., from 47.3 to 46.4 in F1 scores). Combined with labeled training data, ReFF can simultaneously improve both format faithfulness (e.g., from 21.6% in original LLaMA3 to 75.5%) and general quality (e.g., from 47.3 to 61.6 in F1 scores). We further offer an interpretability analysis to explain how ReFF improves both format faithfulness and general quality.

Emerging advancements in large language models (LLMs) show significant potential for enhancing recommendations. However, prompt-based methods often struggle to find ideal prompts without task-specific feedback, while fine-tuning-based methods are hindered by high computational demands and dependence on open-source backbones. To address these challenges, we propose a Reflective Reinforcement Large Language Model (Re2LLM) for session-based recommendation, which refines LLMs to generate and utilize specialized knowledge effectively and efficiently. Specifically, we first devise the Reflective Exploration Module to extract and present knowledge in a form that LLMs can easily process. This module enables LLMs to reflect on their recommendation mistakes and construct a hint knowledge base to rectify them effectively. Next, we design the Reinforcement Utilization Module to train a lightweight retrieval agent that elicits correct LLM reasoning. This module recognizes hints as signals to facilitate LLM recommendations and learns to select appropriate hints from the constructed knowledge base using task-specific feedback efficiently. Lastly, we conduct experiments on real-world datasets and demonstrate the superiority of our Re2LLM over state-of-the-art methods.

Editing videos with textual guidance has garnered popularity due to its streamlined process which mandates users to solely edit the text prompt corresponding to the source video. Recent studies have explored and exploited large-scale text-to-image diffusion models for text-guided video editing, resulting in remarkable video editing capabilities. However, they may still suffer from some limitations such as mislocated objects, incorrect number of objects. Therefore, the controllability of video editing remains a formidable challenge. In this paper, we aim to challenge the above limitations by proposing a Re-Attentional Controllable Video Diffusion Editing (ReAtCo) method. Specially, to align the spatial placement of the target objects with the edited text prompt in a training-free manner, we propose a Re-Attentional Diffusion (RAD) to refocus the cross-attention activation responses between the edited text prompt and the target video during the denoising stage, resulting in a spatially location-aligned and semantically high-fidelity manipulated video. In particular, to faithfully preserve the invariant region content with less border artifacts, we propose an Invariant Region-guided Joint Sampling (IRJS) strategy to mitigate the intrinsic sampling errors w.r.t the invariant regions at each denoising timestep and constrain the generated content to be harmonized with the invariant region content. Experimental results verify that ReAtCo consistently improves the controllability of video diffusion editing and achieves superior video editing performance.

To address the challenges of imbalanced multi-class datasets typically used for rare event detection in critical cyber-physical systems, we propose an optimal, efficient, and adaptable mixed integer programming (MIP) ensemble weighting scheme. Our approach leverages the diverse capabilities of the classifier ensemble on a granular per class basis, while optimizing the weights of classifier-class pairs using elastic net regularization for improved robustness and generalization. Additionally, it seamlessly and optimally selects a predefined number of classifiers from a given set. We evaluate and compare our MIP-based method against six well-established weighting schemes, using representative datasets and suitable metrics, under various ensemble sizes. The experimental results reveal that MIP outperforms all existing approaches, achieving an improvement in balanced accuracy ranging from 0.99% to 7.31%, with an overall average of 4.53% across all datasets and ensemble sizes. Furthermore, it attains an overall average increase of 4.63%, 4.60%, and 4.61% in macro-averaged precision, recall, and F1-score, respectively, while maintaining computational efficiency.

In this paper, we delve into the utilization of the negative momentum technique in constrained minimax games. From an intuitive mechanical standpoint, we introduce a novel framework for momentum buffer updating, which extends the findings of negative momentum from the unconstrained setting to the constrained setting and provides a universal enhancement to the classic game-solver algorithms. Additionally, we provide theoretical guarantees of convergence for our momentum-augmented learning algorithms. We then extend these algorithms to their extensive-form counterparts. Experimental results on both Normal Form Games (NFGs) and Extensive Form Games (EFGs) demonstrate that our momentum techniques can significantly improve algorithm performance, surpassing both their original versions and the SOTA baselines by a large margin.

Automatic Radiology Report Generation (RRG) is an important topic for alleviating the substantial workload of radiologists. Existing RRG approaches rely on supervised regression based on different architectures or additional knowledge injection, while the generated report may not align optimally with radiologists’ preferences. Especially, since the preferences of radiologists are inherently heterogeneous and multi-dimensional, e.g., some may prioritize report fluency, while others emphasize clinical accuracy. To address this problem, we propose a new RRG method via Multi-objective Preference Optimization (MPO) to align the pre-trained RRG model with multiple human preferences, which can be formulated by multi-dimensional reward functions and optimized by multi-objective reinforcement learning (RL). Specifically, we use a preference vector to represent the weight of preferences and use it as a condition for the RRG model. Then, a linearly weighed reward is obtained via a dot product between the preference vector and multi-dimensional reward. Next, the RRG model is optimized to align with the preference vector by optimizing such a reward via RL. In the training stage, we randomly sample diverse preference vectors from the preference space and align the model by optimizing the weighted multi-objective rewards, which leads to an optimal policy on the entire preference space. When inference, our model can generate reports aligned with specific preferences without further fine-tuning. Extensive experiments on two public datasets show the proposed method can generate reports that cater to different preferences in a single model and achieve state-of-the-art performance.

The Dynamic Retrieval Augmented Generation (RAG) paradigm actively decides when and what to retrieve during the text generation process of Large Language Models (LLMs). However, current dynamic RAG methods fall short in both aspects: identifying the optimal moment to activate the retrieval module and crafting the appropriate query once retrieval is triggered. To overcome these limitations, we introduce an approach, namely, RaDIO, Real-Time Hallucination Detection with Contextual Index Optimized query formulation for dynamic RAG. The approach is specifically designed to make decisions on when and what to retrieve based on the LLM’s real-time information needs during the text generation process. We evaluate RaDIO along with existing methods comprehensively over several knowledge-intensive generation datasets. Experimental results show that RaDIO achieves superior performance on all tasks, demonstrating the effectiveness of our work.

Multimodal fake news detection aims to automatically identify real or fake news, thereby mitigating the adverse effects caused by such misinformation. Although prevailing approaches have demonstrated their effectiveness, challenges persist in cross-modal feature fusion and refinement for classification. To address this, we present a residual-aware compensation network with multi-granularity constraints (RaCMC) for fake news detection, that aims to sufficiently interact and fuse cross-modal features while amplifying the differences between real and fake news. First, a multiscale residual-aware compensation module is designed to interact and fuse features at different scales, and ensure both the consistency and exclusivity of feature interaction, thus acquiring high-quality features. Second, a multi-granularity constraints module is implemented to limit the distribution of both the news overall and the image-text pairs within the news, thus amplifying the differences between real and fake news at the news and feature levels. Finally, a dominant feature fusion reasoning module is developed to comprehensively evaluate news authenticity from the perspectives of both consistency and inconsistency. Experiments on three public datasets, including Weibo17, Politifact and GossipCop, reveal the superiority of the proposed method.

Human life is filled with articulated objects. Previous works for estimating the pose of category-level articulated objects rely on costly 3D point clouds or RGB-D images. In this paper, our goal is to estimate category-level articulation poses from a single RGB image, where we propose R2-Art, a novel category-level Articulation pose estimation framework from a single RGB image and a cascade Render strategy. Given an RGB image as input, R2-Art estimates per-part 6D pose for the articulation. Specifically, we design parallel regression branches tailored to generate camera-to-root translation and rotation. Using the predicted joint states, we perform PC prior transformation and deformation with a joint-centric modeling approach. For further refinement, a cascade render strategy is proposed for projecting the 3D deformed prior onto the 2D mask. Extensive experiments are provided to validate our R2-Art on various datasets ranging from synthetic datasets to real-world scenarios, demonstrating the superior performance and robustness of the R2-Art. We believe that this work has the potential to be applied in many fields including robotics, embodied intelligence, and augmented reality.

Enabling object detectors to recognize out-of-distribution (OOD) objects is vital for building reliable systems. A primary obstacle stems from the fact that models frequently do not receive supervisory signals from unfamiliar data, leading to overly confident predictions regarding OOD objects. Despite previous progress that estimates OOD uncertainty based on the detection model and in-distribution (ID) samples, we explore using pre-trained vision-language representations for object-level OOD detection. We first discuss the limitations of applying image-level CLIP-based OOD detection methods to object-level scenarios. Building upon these insights, we propose RUNA, a novel framework that leverages a dual encoder architecture to capture rich contextual information and employs a regional uncertainty alignment mechanism to distinguish ID from OOD objects effectively. We introduce a few-shot fine-tuning approach that aligns region-level semantic representations to further improve the model's capability to discriminate between similar objects. Our experiments show that RUNA substantially surpasses state-of-the-art methods in object-level OOD detection, particularly in challenging scenarios with diverse and complex object instances.

Large language models (LLMs) and small language models (SLMs) are being adopted at remarkable speed, although their safety still remains a serious concern. With the advent of multilingual S/LLMs, the question now becomes a matter of scale: can we expand multilingual safety evaluations of these models with the same velocity at which they are deployed? To this end, we introduce RTP-LX, a human-transcreated and human-annotated corpus of toxic prompts and outputs in 28 languages. RTP-LX follows participatory design practices, and a portion of the corpus is especially designed to detect culturally-specific toxic language. We evaluate 10 S/LLMs on their ability to detect toxic content in a culturally-sensitive, multilingual scenario. We find that, although they typically score acceptably in terms of accuracy, they have low agreement with human judges when scoring holistically the toxicity of a prompt; and have difficulty discerning harm in context-dependent scenarios, particularly with subtle-yet-harmful content (e.g. microaggressions, bias). We release this dataset to contribute to further reduce harmful uses of these models and improve their safe deployment.

Learning-based multi-view stereo methods aim to predict depth maps for reconstructing dense point clouds. These methods rely on regularization to reduce redundancy in the cost volume. However, existing methods have limitations: CNN-based regularization is restricted to local receptive fields, while Transformer-based regularization struggles with handling depth discontinuities. These limitations often result in inaccurate depth maps with significant noise, particularly noticeable in the boundary and background regions. In this paper, we propose a Recurrent Regularization Transformer for Multi-View Stereo (RRT-MVS), which addresses these limitations by regularizing the cost volume separately for depth and spatial dimensions. Specifically, we introduce Recurrent Self-Attention (R-SA) to aggregate global matching costs within and across the cost maps and filter out noisy feature correlations. Additionally, we present Depth Residual Attention (DRA) to aggregate depth correlations within the cost volume and a Positional Adapter (PA) to enhance 3D positional awareness in each 2D cost map, further augmenting the effectiveness of R-SA. Experimental results demonstrate that RRT-MVS achieves state-of-the-art performance on the DTU and Tanks-and-Temples datasets. Notably, RRT-MVS ranks first on both the Tanks-and-Temples intermediate and advanced benchmarks among all published methods.

Adversarial attack and defense have been extensively explored in classification tasks, but their study in semantic segmentation remains limited. Moreover, current attacks fail to act as strong underlying attacks for adversarial training (AT), making it difficult to achieve segmentation robustness against strong attacks. In this paper, we present RP-PGD, a novel Region-and-Prototype based Projected Gradient Descent attack tailored to fool segmentation models. In particular, we propose a region-based attack, which leverages a spatial-temporal way to separate the pixels into three disjoint regions, and highlights the attack on the crucial True Region and Boundary Region. Moreover, we introduce a prototype-based attack to disrupt the feature space, further enhancing the attack capability. To boost the robustness of segmentation models, we inject adversaries generated by RP-PGD into the clean data and perform AT. Extensive experiments on multiple datasets showcase that RP-PGD generates adversaries with faster convergence and stronger attack effectiveness, surpassing state-of-the-art attacks by a large margin. Consequently, RP-PGD serves as a strong underlying attack for segmentation models to perform AT, assisting them in defending against a variety of strong attacks without incurring additional computational costs during inference.

Mathematical reasoning ability objectively reflects a language model's understanding of implicit knowledge in contexts, with logic being a prerequisite for exploring, articulating and establishing effective reasoning. Large language models (LLMs) have shown great potential in complex reasoning tasks represented by mathematical reasoning. However, existing mathematical datasets either focus on commonsense reasoning, assessing the model's knowledge application ability, or arithmetic problems with fixed calculation rules, evaluating the model's rapid learning capability. There is a lack of datasets that require solving problems solely through logical reasoning. As a result, the performance of LLMs in accurately understanding the implicit logical relationships in problems and deriving conclusions based solely on given conditions is hindered. To address this challenge, we construct a dataset specifically for multiple step reasoning tasks: Reasoning-Math (RMath). This dataset focuses on evaluating logical reasoning abilities with mathematical reasoning problems, covering typical problem types, including direct reasoning problems, hypothetical reasoning problems, and nested reasoning problems. Additionally, we design a standardized annotation scheme that transforms natural language descriptions of conditions into formal propositions. Other annotation contents include problem categories, proposition truth values, and proposition relationship types. This not only reduces biases caused by semantic misunderstandings during problem-solving, but also facilitates the incorporation of theoretically grounded logical reasoning methods to enhance reasoning abilities. Furthermore, we propose a normalization problem-solving framework based on propositional logic for RMath and design the problem-solving process for prompt tuning to guide LLMs to absorb mathematical logical theories and improving reasoning abilities. Finally, we evaluate RMath on several popular LLMs and present the corresponding results.

LLM-powered personalization agent systems employ Large Language Models (LLMs) to predict users’ behavior from their past activities. However, their effectiveness often hinges on the ability to effectively leverage extensive, long user historical data due to its inherent noise and length of such data. Existing pre-trained LLMs may generate summaries that are concise but lack the necessary context for downstream tasks, hindering their utility in personalization systems. To address these challenges, we introduce Reinforcement Learning from Prediction Feedback (RLPF). RLPF fine-tunes LLMs to generate concise, human-readable user summaries that are optimized for downstream task performance. By maximizing the usefulness of the generated summaries, RLPF effectively distills extensive user history data while preserving essential information for downstream tasks. Our empirical evaluation demonstrates significant improvements in both extrinsic downstream task utility and intrinsic summary quality, surpassing baseline methods by up to 22% and achieving an up to 84.59% win rate on Factuality, Abstractiveness, and Readability. RLPF also achieves a remarkable 74% reduction while improving performance on 16 out of 19 unseen tasks and/or datasets, showcasing its generalizability. This approach offers a promising solution for enhancing LLM personalization by effectively transforming long, noisy user histories into informative and human-readable representations.

We present RLLTE: a long-term evolution, extremely modular, and open-source framework for reinforcement learning (RL) research and application. Beyond delivering top-notch algorithm implementations, RLLTE also serves as a toolkit for developing algorithms. More specifically, RLLTE decouples the RL algorithms completely from the exploitation-exploration perspective, providing a large number of components to accelerate algorithm development and evolution. In particular, RLLTE is the first RL framework to build a comprehensive ecosystem, which includes model training, evaluation, deployment, benchmark hub, and large language model (LLM)-empowered copilot. RLLTE is expected to set standards for RL engineering practice and be highly stimulative for industry and academia. Our documentation, examples, and source code are available at https://github.com/RLE-Foundation/rllte.

Low-rank adaptation (LoRA) has become the dominant method for parameter-efficient LLM fine-tuning, with LoRA-based quantization error compensation (LQEC) emerging as a powerful tool for recovering accuracy in compressed LLMs. However, LQEC has underperformed in sub-4-bit scenarios, with no prior investigation into understanding this limitation. We propose RILQ (Rank-Insensitive LoRA-based Quantization Error Compensation) to boost 2-bit LLM accuracy. Based on rank analysis revealing model-wise activation discrepancy loss's rank-insensitive nature, RILQ employs this loss to adjust adapters cooperatively across layers, enabling robust error compensation with low-rank adapters. Evaluations on LLaMA-2 and LLaMA-3 demonstrate RILQ's consistent improvements in 2-bit quantized inference across various state-of-the-art quantizers and enhanced accuracy in task-specific fine-tuning. RILQ maintains computational efficiency comparable to existing LoRA methods, enabling adapter-merged weight-quantized LLM inference with significantly enhanced accuracy, making it a promising approach for boosting 2-bit LLM performance.

Masked point modeling methods have recently achieved great success in self-supervised learning for point cloud data. However, these methods are sensitive to rotations and often exhibit sharp performance drops when encountering rotational variations. In this paper, we propose a novel Rotation-Invariant Masked AutoEncoders (RI-MAE) to address two major challenges: 1) achieving rotation-invariant latent representations, and 2) facilitating self-supervised reconstruction in a rotation-invariant manner. For the first challenge, we introduce RI-Transformer, which features disentangled geometry content, rotation-invariant relative orientation and position embedding mechanisms for constructing rotation-invariant point cloud latent space. For the second challenge, a novel dual-branch student-teacher architecture is devised. It enables the self-supervised learning via the reconstruction of masked patches within the learned rotation-invariant latent space. Each branch is based on an RI-Transformer, and they are connected with an additional RI-Transformer predictor. The teacher encodes all point patches, while the student solely encodes unmasked ones. Finally, the predictor predicts the latent features of the masked patches using the output latent embeddings from the student, supervised by the outputs from the teacher. Extensive experiments demonstrate that our method is robust to rotations, achieving the state-of-the-art performance on various downstream tasks.

Although diffusion models can generate high-quality human images, their applications are limited by the instability in generating hands with correct structures. In this paper, we introduce RHanDS, a conditional diffusion-based framework designed to refine malformed hands by utilizing decoupled structure and style guidance. The hand mesh reconstructed from the malformed hand offers structure guidance for correcting the structure of the hand, while the malformed hand itself provides style guidance for preserving the style of the hand. To alleviate the mutual interference between style and structure guidance, we introduce a two-stage training strategy and build a series of multi-style hand datasets. In the first stage, we use paired hand images for training to ensure stylistic consistency in hand refining. In the second stage, various hand images generated based on human meshes are used for training, enabling the model to gain control over the hand structure. Experimental results demonstrate that RHanDS can effectively refine hand structure while preserving consistency in hand style.