AAAI 2025 论文列表

Semantic segmentation and 3D reconstruction are two fundamental tasks in remote sensing, typically treated as separate or loosely coupled tasks. Despite attempts to integrate them into a unified network, the constraints between the two heterogeneous tasks are not explicitly modeled, since the pioneering studies either utilize a loosely coupled parallel structure or engage in only implicit interactions, failing to capture the inherent connections. In this work, we explore the connections between the two tasks and propose a new network that imposes semantic constraints on the stereo matching task, both implicitly and explicitly. Implicitly, we transform the traditional parallel structure to a new cascade structure termed Semantic-Guided Cascade structure, where the deep features enriched with semantic information are utilized for the computation of initial disparity maps, enhancing semantic guidance. Explicitly, we propose a Semantic Selective Refinement (SSR) module and a Left-Right Semantic Consistency (LRSC) module. The SSR refines the initial disparity map under the guidance of the semantic map. The LRSC ensures semantic consistency between two views via reducing the semantic divergence after transforming the semantic map from one view to the other using the disparity map. Experiments on the US3D and WHU datasets demonstrate that our method achieves state-of-the-art performance for both semantic segmentation and stereo matching.

Multi-view clustering (MVC), especially contrastive MVC, has demonstrated promising potential in many fields and practical scenarios. However, existing contrastive MVC methods still ignore the reliability of clustering results and the impact of false negative pairs, which limits the application of methods in critical security areas. To solve the above challenges, we propose a Self-supervised Trusted Contrastive Multi-view Clustering with Uncertainty Refined (STCMC-UR) method, which integrates clustering results and uncertainty learning to guide the self-supervised contrastive learning (CL). First, the belief of a specific view is generated in the evidence generation module. Afterwards, the belief mass and uncertainty of each view are learned using the Dirichlet distribution and we fuse multiple views with the Dempster-Shafer theory to generate the final clustering result and the uncertainty of the view. Then, the view weight is further quantified to adjust the belief of each view. Different from existing methods, with the clustering result and uncertainty generated by the fusion, we design a feature-level uncertainty-refined self-supervised CL module, where the pseudo-label is selectively employed in each iteration to conduct more accurate CL. As a result, the modules are mutually beneficial, which is conducive to more effective feature learning and clustering structure discovery, and more accurate learning results are obtained. Extensive experiments on five datasets show that the proposed method has significant improvements in effectiveness compared with the latest methods.

Missing values in multivariate time series data can harm machine learning performance and introduce bias. These gaps arise from sensor malfunctions, blackouts, and human error and are typically addressed by data imputation. Previous work has tackled the imputation of missing data in random, complete blackouts and forecasting scenarios. The current paper addresses a more general missing pattern, which we call "partial blackout," where a subset of features is missing for consecutive time steps. We introduce a two-stage imputation process using self-attention and diffusion processes to model feature and temporal correlations. Notably, our model effectively handles missing data during training, enhancing adaptability and ensuring reliable imputation and performance, even with incomplete datasets. Our experiments on benchmark and two real-world time series datasets demonstrate that our model outperforms the state-of-the-art in partial blackout scenarios and shows better scalability.

Text readability assessment involves categorizing texts based on readers' comprehension levels. Hybrid automatic readability assessment (ARA) models, combining deep and linguistic features, have recently attracted rising attention due to their impressive performance. However, existing hybrid ARA models generally ignore the specific-intrinsic information of deep and linguistic representations, and cannot fully explore their common-intrinsic information. In this paper, we introduce a self-supervised collaborative information bottleneck (SCIB) module for ARA to address these issues. Specifically, we collaboratively consider both specific-intrinsic and common-intrinsic information of the linguistic representation and various levels of deep representations including the document-, sentence- and word-level deep representations, and yield their refined representations via a self-supervised information bottleneck scheme. Extensive experiments are conducted on four English and two Chinese corpora to demonstrate the effectiveness of the proposed model. Experimental results show that the proposed model outperforms state-of-the-art models in terms of four important evaluation metrics, and the suggested SCIB module can effectively capture the specific- and common-intrinsic information.

Unmanned Aerial Vehicle Object Detection (UAVOD) presents unique challenges due to varying altitudes, dynamic backgrounds, and the small size of objects. Traditional detection methods often struggle with these challenges, as they typically rely on visual feature only and fail to extract the semantic relations between the objects. To address these limitations, we propose a novel approach named Self-Prompting Analogical Reasoning (SPAR). Our method utilizes the vision-language model (CLIP) to generate context-aware prompts based on image feature, providing rich semantic information that guides analogical reasoning. SPAR includes two main modules: self-prompting and analogical reasoning. Self-prompting module based on learnable description and CLIP-text encoder generates context-aware prompt by combining specific image feature; then an objectness prompt score map is produced by computing the similarity between pixel-level features and context-aware prompt. With this score map, multi-scale image features are enhanced and pixel-level features are chosen for graph construction. While for analogical reasoning module, graph nodes consists of category-level prompt nodes and pixel-level image feature nodes. Analogical inference is based graph convolution. Under the guidance of category-level nodes, different-scale object features have been enhanced, which helps achieve more accurate detection of challenging objects. Extensive experiments illustrate that SPAR outperforms traditional methods, offering a more robust and accurate solution for UAVOD.

Signed Graph Neural Networks (SGNNs) have been shown to be effective in analyzing complex patterns in real-world situations where positive and negative links coexist. However, SGNN models suffer from poor explainability, which limit their adoptions in critical scenarios that require understanding the rationale behind predictions. To the best of our knowledge, there is currently no research work on the explainability of the SGNN models. Our goal is to address the explainability of decision-making for the downstream task of link sign prediction specific to signed graph neural networks. Since post-hoc explanations are not derived directly from the models, they may be biased and misrepresent the true explanations. Therefore, in this paper we introduce a Self-Explainable Signed Graph transformer (SE-SGformer) framework, which can not only outputs explainable information while ensuring high prediction accuracy. Specifically, we propose a new Transformer architecture for signed graphs and theoretically demonstrate that using positional encoding based on signed random walks has greater expressive power than current SGNN methods and other positional encoding graph Transformer-based approaches. We construct a novel explainable decision process by discovering the K-nearest (farthest) positive (negative) neighbors of a node to replace the neural network-based decoder for predicting edge signs. These K positive (negative) neighbors represent crucial information about the formation of positive (negative) edges between nodes and thus can serve as important explanatory information in the decision-making process. We conducted experiments on several real-world datasets to validate the effectiveness of SE-SGformer, which outperforms the state-of-the-art methods by improving 2.2% prediction accuracy and 73.1% explainablity accuracy in the best-case scenario.

Iterative preference optimization has recently become one of the de-facto training paradigms for large language models (LLMs), but the performance is still underwhelming due to too much noisy preference data yielded in the loop. To combat this issue, we present an Uncertainty-enhanced Preference Optimization (UPO) framework to make the LLM self-evolve with reliable feedback. The key idea is mitigating the noisy preference pairs derived from the current policy and reward models by performing pair-wise uncertainty estimation and judiciously reliable feedback sampling. To reach this goal, we thus introduce an estimator model, which incorporates Monte Carlo (MC) dropout in Bayesian neural network (BNN) to perform uncertainty estimation for the batch of preference pairs. Compared to the existing methods that directly filter generated responses based on the reward score, the estimator focuses on the model uncertainty in a pair-wise manner and effectively bypasses the confirmation bias problem of the reward model. Additionally, we also propose an uncertainty-enhanced self-evolution algorithm to better improve the LLM robustly align with these reliable feedback data. Extensive experiments over multiple benchmarks demonstrate our framework substantially improves the performance of iterative preference optimization.

Language-conditioned robotic manipulation in unstructured environments presents significant challenges for intelligent robotic systems. However, due to partial observation or imprecise action prediction, failure may be unavoidable for learned policies. Moreover, operational failures can lead to the robotic arm entering an untrained state, potentially causing destructive results. Consequently, the ability to detect and self-correct failures is crucial for the development of practical robotic systems. To address this challenge, we propose a foresight-driven failure detection and self-correction module for robot manipulation. By leveraging 3D Gaussian Splatting, we represent the current scene with multiple Gaussians. Subsequently, we train a prediction network to forecast the Gaussian representation of future scenes conditioned on planned actions. Failure is detected when the predicted future significantly deviates from the real observation after action execution. In such cases, the end-effector rolls back to the previous action to avoid an untrained state. Integrating this approach with the PerACT framework, we develop a self-correcting robot manipulation policy. Evaluations on ten RLBench tasks with 166 variations demonstrate the superior performance of the proposed method, which outperforms state-of-the-art methods by 12.0% success rate on average.

Flow matching has emerged as a promising framework for training generative models, demonstrating impressive empirical performance while offering relative ease of training compared to diffusion-based models. However, this method still requires numerous function evaluations in the sampling process. To address these limitations, we introduce a self-corrected flow distillation method that effectively integrates consistency models and adversarial training within the flow-matching framework. This work is a pioneer in achieving consistent generation quality in both few-step and one-step sampling. Our extensive experiments validate the effectiveness of our method, yielding superior results both quantitatively and qualitatively on CelebA-HQ and zero-shot benchmarks on the COCO dataset.

Spiking Neural Networks (SNNs) are promising for low-power computation due to their event-driven mechanism but often suffer from lower accuracy compared to Artificial Neural Networks (ANNs). ANN-to-SNN knowledge distillation can improve SNN performance, but previous methods either focus solely on label information, missing valuable intermediate layer features, or use a layer-wise approach that neglects spatial and temporal semantic inconsistencies, leading to performance degradation. To address these limitations, we propose a novel method called self-attentive spatio-temporal calibration (SASTC). SASTC uses self-attention to identify semantically aligned layer pairs between ANN and SNN, both spatially and temporally. This enables the autonomous transfer of relevant semantic information. Extensive experiments show that SASTC outperforms existing methods, effectively solving the mismatching problem. Superior accuracy results include 95.12% on CIFAR-10, 79.40% on CIFAR-100 with 2 time steps, and 68.69% on ImageNet with 4 time steps for static datasets, and 97.92% on DVS-Gesture and 83.60% on DVS-CIFAR10 for neuromorphic datasets. This marks the first time SNNs have outperformed ANNs on both CIFAR-10 and CIFAR-100, shedding the new light on the potential applications of SNNs.

Pre-trained Vision Mamba~(Vim) models have demonstrated exceptional performance across various computer vision tasks in a computationally efficient manner, attributed to their unique design of selective state space models. To further extend their applicability to diverse downstream vision tasks, Vim models can be adapted using the efficient fine-tuning technique known as visual prompting. However, existing visual prompting methods are predominantly tailored for Vision Transformer (ViT)-based models that leverage global attention, neglecting the distinctive sequential token-wise compression and propagation characteristics of Vim. Specifically, existing prompt tokens prefixed to the sequence are insufficient to effectively activate the input and forget gates across the entire sequence, hindering the extraction and propagation of discriminative information. To address this limitation, we introduce a novel Selective Visual Prompting (SVP) method specifically for the efficient fine-tuning of Vim. To prevent the loss of discriminative information during state space propagation, SVP employs lightweight selective prompters for token-wise prompt generation, ensuring adaptive activation of the update and forget gates within Mamba blocks to promote discriminative information propagation. Moreover, considering that Vim propagates both shared cross-layer information and specific inner-layer information, we further refine SVP with a dual-path structure: Cross-Prompting and Inner-Prompting. Cross-Prompting utilizes shared parameters across layers, while Inner-Prompting employs distinct parameters, promoting the propagation of both shared and specific information, respectively. Extensive experimental results on various large-scale benchmarks demonstrate that our proposed SVP significantly outperforms state-of-the-art methods.

We consider the finite-horizon offline reinforcement learning (RL) setting, and are motivated by the challenge of learning the policy at any step h in dynamic programming (DP) algorithms. To learn this, it is sufficient to evaluate the treatment effect of deviating from the behavioral policy at step h after having optimized the policy for all future steps. Since the policy at any step can affect next-state distributions, the related distributional shift challenges can make this problem far more statistically hard than estimating such treatment effects in the stochastic contextual bandit setting. However, the hardness of many real-world RL instances lies between the two regimes. We develop a flexible and general method called selective uncertainty propagation for confidence interval construction that adapts to the hardness of the associated distribution shift challenges. We show benefits of our approach on toy environments and demonstrate the benefits of these techniques for offline policy learning.

This paper explores Machine Unlearning (MU), an emerging field that is gaining increased attention due to concerns about neural models unintentionally remembering personal or sensitive information. We present SeUL, a novel method that enables selective and fine-grained unlearning for language models. Unlike previous work that employs a fully reversed training objective in unlearning, SeUL minimizes the negative impact on the capability of language models, particularly in terms of generation. Furthermore, we introduce two innovative evaluation metrics, sensitive extraction likelihood (S-EL) and sensitive memorization accuracy (S-MA), specifically designed to assess the effectiveness of forgetting sensitive information. In support of the unlearning framework, we propose efficient automatic online and offline sensitive span annotation methods. The online selection method, based on language probability scores, ensures computational efficiency, while the offline annotation involves a two-stage LLM-based process for robust verification. In summary, this paper contributes a novel selective unlearning method (SeUL), introduces specialized evaluation metrics (S-EL and S-MA) for assessing sensitive information forgetting, and proposes automatic online and offline sensitive span annotation methods to support the overall unlearning framework and evaluation.

This paper presents SAGA (Segment Any 3D GAussians), a highly efficient 3D promptable segmentation method based on 3D Gaussian Splatting (3D-GS). Given 2D visual prompts as input, SAGA can segment the corresponding 3D target represented by 3D Gaussians within 4 ms. This is achieved by attaching a scale-gated affinity feature to each 3D Gaussian to endow it a new property towards multi-granularity segmentation. Specifically, a scale-aware contrastive training strategy is proposed for the scale-gated affinity feature learning. It 1) distills the segmentation capability of the Segment Anything Model (SAM) from 2D masks into the affinity features and 2) employs a soft scale gate mechanism to deal with multi-granularity ambiguity in 3D segmentation through adjusting the magnitude of each feature channel according to a specified 3D physical scale. Evaluations demonstrate that SAGA achieves real-time multi-granularity segmentation with quality comparable to state-of-the-art methods. As one of the first methods addressing promptable segmentation in 3D-GS, the simplicity and effectiveness of SAGA pave the way for future advancements in this field.

Face parsing refers to the semantic segmentation of human faces into key facial regions such as eyes, nose, hair, etc. It serves as a prerequisite for various advanced applications, including face editing, face swapping, and facial makeup, which often require segmentation masks for classes like eyeglasses, hats, earrings, and necklaces. These infrequently occurring classes are called long-tail classes, which are overshadowed by more frequently occurring classes known as head classes. Existing methods, primarily CNN-based, tend to be dominated by head classes during training, resulting in suboptimal representation for long-tail classes. Previous works have largely overlooked the problem of poor segmentation performance of long-tail classes. To address this issue, we propose SegFace, a simple and efficient approach that uses a lightweight transformer-based model which utilizes learnable class-specific tokens. The transformer decoder leverages class-specific tokens, allowing each token to focus on its corresponding class, thereby enabling independent modeling of each class. The proposed approach improves the performance of long-tail classes, thereby boosting overall performance. To the best of our knowledge, SegFace is the first work to employ transformer models for face parsing. Moreover, our approach can be adapted for low-compute edge devices, achieving 95.96 FPS. We conduct extensive experiments demonstrating that SegFace significantly outperforms previous state-of-the-art models, achieving a mean F1 score of 88.96 (+2.82) on the CelebAMask-HQ dataset and 93.03 (+0.65) on the LaPa dataset.

LIDAR-based 3D object detection and semantic segmentation are critical tasks in 3D scene understanding. Traditional detection and segmentation methods supervise their models through bounding box labels and semantic mask labels. However, these two independent labels inherently contain significant redundancy. This paper aims to eliminate the redundancy by supervising 3D object detection using only semantic labels. However, the challenge arises due to the incomplete geometry structure and boundary ambiguity of point cloud instances, leading to inaccurate pseudo-labels and poor detection results. To address these challenges, we propose a novel method, named Seg2Box. We first introduce a Multi-Frame Multi-Scale Clustering (MFMS-C) module, which leverages the spatio-temporal consistency of point clouds to generate accurate box-level pseudo-labels. Additionally, the Semantic-Guiding Iterative-Mining Self-Training (SGIM-ST) module is proposed to enhance the performance by progressively refining the pseudo-labels and mining the instances without generating pseudo-labels. Experiments on the Waymo Open Dataset and nuScenes Dataset show that our method significantly outperforms other competitive methods by 23.7% and 10.3% in mAP, respectively. The results demonstrate the great label-efficient potential and advancement of our method.

Face-based Voice Conversion (FVC) is a novel task that leverages facial images to generate the target speaker's voice style. Previous work has two shortcomings: (1) suffering from obtaining facial embeddings that are well-aligned with the speaker's voice identity information, and (2) inadequacy in decoupling content and speaker identity information from the audio input. To address these issues, we present a novel FVC method, Identity-Disentanglement Face-based Voice Conversion (ID-FaceVC), which overcomes the above two limitations. More precisely, we propose an Identity-Aware Query-based Contrastive Learning (IAQ-CL) module to extract speaker-specific facial features, and a Mutual Information-based Dual Decoupling (MIDD) module to purify content features from audio, ensuring clear and high-quality voice conversion. Besides, unlike prior works, our method can accept either audio or text inputs, offering controllable speech generation with adjustable emotional tone and speed. Extensive experiments demonstrate that ID-FaceVC achieves state-of-the-art performance across various metrics, with qualitative and user study results confirming its effectiveness in naturalness, similarity, and diversity.

Multimodal Recommendation Systems (MRSs) boost traditional user-item interaction-based methods by incorporating multimodal information. However, existing methods ignore the inherent noise brought by (1) noisy semantic priors in multimodal content, and (2) noisy user interactions in history records, therefore diminishing model performance. To fill this gap, we propose to denoise MRSs by jointly EValuating structure Effectiveness and mitigating Noisy links (EVEN). Firstly, for semantic prior noise in multimodal content, EVEN builds item homogeneous consistency and denoises it by evaluating behavior-driven confidence. Secondly, for noise in user interactions, EVEN updates user feedback by denoising observed interactions following implicit contribution evaluation of high-order representations. Thirdly, EVEN performs cross-modal alignment through self-guided structure learning, reinforcing task-specific inter-modal dependency modeling and cross-modal fusion. Through extensive experiments on three widely-used datasets, EVEN achieves an average improvement of 8.95% and 5.90% in recommendation accuracy compared with LGMRec and FREEDOM, respectively, without extending the total training time.

Entrusted with the goal of pixel-level object classification, the semantic segmentation networks entails the laborious preparation of pixel-level annotation masks. To obtain pixel-level annotation masks for a given class without human efforts, recent few works have proposed to generate pairs of images and annotation masks by employing image and text relationships modeled by text-to-image generative models, especially Stable Diffusion. However, these works do not fully exploit the capability of text-guided Diffusion models and thus require a pre-trained segmentation network, careful text prompt tuning, or the training of a segmentation network to generate final annotation masks. In this work, we take a closer look at attention mechanisms of Stable Diffusion, from which we draw connections with classical seeded segmentation approaches. In particular, we show that cross-attention alone provides very coarse object localization, which however can provide initial seeds. Then, akin to region expansion in seeded segmentation, we utilize the semantic-correspondence-modeling capability of self-attention to iteratively spread the attention to the whole class from the seeds using multi-scale self-attention maps. We also observe that a simple-text-guided synthetic image often has a uniform background, which is easier to find correspondences, compared to complex-structured objects. Thus, we further refine a mask using a more accurate background mask. Our proposed method, dubbed SeeDiff, generates high-quality masks off-the-shelf from Stable Diffusion, without additional training procedure, prompt tuning, or a pre-trained segmentation network.