Singapore

Large Vision-Language Models (LVLMs) enhance performance on vision-language tasks by integrating visual features from pre-trained vision encoders into large language models (LLMs). However, the large number of visual tokens introduces significant computational overhead. Existing token pruning methods either perform global selection via [CLS]-based attention in the vision encode or prune within LLM decoding layers. These approaches face two key challenges: (1) [CLS]-based attention primarily focuses on visually salient regions across the entire image, often overlooking semantically important tokens essential for reasoning; and (2) strong positional bias in the shallow decoder layers causes the model to favor later-positioned tokens, while neglecting earlier ones that may carry critical reasoning cues. To address these issues, we propose PosPrune, a training-free, two-stage visual token pruning framework. At the vision encoder, we introduce an Asymmetric Region-aware Pruning (ARP) strategy that retains more tokens in semantically rich regions while discarding more tokens from semantically less informative regions, thus preserving spatial diversity and task-relevant details. In the LLM decoding stage, we find that the positional bias in shallow layers is primarily driven by model architecture rather than task semantics. Based on this insight, we propose a novel Positional Bias Correction (PBC) mechanism to mitigate this bias. To further reduce redundancy, we apply Maximal Marginal Relevance (MMR) to select tokens that best balance textual relevance and diversity. Extensive experiments on various LVLMs and benchmarks demonstrate the general effectiveness of our approach. Notably, when applied to LLaVA-1.5-7B, PosPrune achieves a reduction of 85% in FLOPs while preserving 98.5% of the original performance.

AAAI 2026

PosPrune: Visual Token Pruning with Positional Bias Correction for Efficient Large Vision-Language Models

large vision models

multi-modal vision

language and vision

We are pleased to announce the Fortieth AAAI Conference on Artificial Intelligence (AAAI-26), which will be held in Singapore EXPO from January 20 to January 27, 2026.

The purpose of the AAAI conference series is to promote research in Artificial Intelligence (AI) and foster scientific exchange between researchers, practitioners, scientists, students, and engineers across the entirety of AI and its affiliated disciplines. AAAI-26 will feature technical paper presentations, special tracks, invited speakers, workshops, tutorials, poster sessions, senior member presentations, competitions, and exhibit programs, and a range of other activities to be announced.<br><br>

To access this event page, you need to log in with the **email address you registered with**. <br>Access credentials will be sent to your email from Underline -  subject line "Welcome to AAAI 2026". Please be sure to check your spam email folder if you do not see an email confirmation right away.

Please log in

To access this event page, you are required to register.
Please complete your registration to continue.

We recommend reading [**the registration information**](https://aaai.org/conference/aaai/aaai-26/registration/) first.

**Online Registration Form**: https://aaai.getregistered.net/conference-2026 

Registration Required

We are pleased to announce the Fortieth AAAI Conference on Artificial Intelligence (AAAI-26), which will be held in Singapore EXPO from January 20 to January 27, 2026.

poster

Cloud removal (CR) in remote sensing imagery is a critical yet challenging task due to complex cloud patterns and diverse underlying ground structures. Despite recent progress in generative models such as diffusion models, CR remains limited by its inadequate capability to perceive and reconstruct structured information beneath cloud-covered areas. In this work, we propose a Visibility-guided Semantic Estimation and Reconstruction network for cloud removal (VISER-CR), which reformulates CR as a structure-guided completion problem. Specifically, VISER-CR explicitly models cloud interference via spatial masking, encouraging the model to reason beyond pixel-level appearance and enhance scene-level structural understanding. Moreover, to further improve the representation of structural information, we introduce Patch Saliency Encoding, a self-guided mechanism that implicitly models structural alignment among patches, significantly enhancing clustering consistency and semantic separability in the latent space. This adaptive mechanism guides the network to focus on learning and reconstructing structurally important regions, thereby reducing redundancy and improving overall cloud removal performance. Extensive experiments on multiple benchmark datasets demonstrate the superior effectiveness of our method.

Revealing the Invisible: Latent Structure Modeling for Semantically Consistent Cloud Removal

Robot navigation in dense crowds requires understanding social cues that humans naturally use, yet existing methods struggle with real-world complexity. We investigate two questions: (1) Where do pedestrians look when navigating crowds? and (2) Can eye tracking improve robot navigation? To answer, we introduce GazeNav, an egocentric dataset collected via wearable eye trackers, featuring synchronized video, gaze, and trajectories in crowded environments. Analysis reveals that the gaze of pedestrians is closely related to the semantic presence and movement of other individuals, exhibiting distinct attention patterns across navigation behaviors. Building on this, we propose Gaze2Nav, a modular framework that first predicts human gaze to infer socially salient pedestrians, then incorporates the semantic attention into motion planning alongside visual inputs. Our method achieves 87.6% salient pedestrian prediction accuracy and reduces trajectory error by 15.4% over state-of-the-art baselines. By aligning with human gaze, our framework improves both performance and interpretability, advancing toward human-like, socially intelligent robot navigation.

Learning from Human Gaze: Human-like Robot Social Navigation in Dense Crowds

Large-scale EEG foundation models have shown strong generalization across a range of downstream tasks, but their training remains resource-intensive due to the volume and variable quality of EEG data. In this work, we introduce EEG-D², a data distillation framework that enables more efficient pre-training by selectively removing noisy and redundant samples from large EEG datasets. EEG-D² begins by encoding EEG segments into compact latent representations using a self-supervised autoencoder, allowing sample selection to be performed efficiently and with reduced sensitivity to noise. Based on these representations, EEG-D² filters out outliers and minimizes redundancy, resulting in a smaller yet informative subset that retains the diversity essential for effective foundation model training. Through extensive experiments, we demonstrate that training on only 5 percent of a 2,500-hour dataset curated with EEG-D² yields performance comparable to, and in some cases better than, training on the full dataset across multiple downstream tasks. To our knowledge, this is the first systematic study of pre-training data distillation in the context of EEG foundation models. EEG-D² provides a scalable and practical path toward more effective and efficient physiological foundation modeling.

EEG-DLite: Dataset Distillation for Efficient Large EEG Model Training

3D Gaussian Splatting (3DGS) has emerged as a mainstream solution for real-time rendering and high-fidelity novel view synthesis. Building on this foundation, methods based on Textured Gaussians further improve the expression ability by incorporating explicit texture mapping into Gaussians. However, their reliance on fixed texture resolution often results in noticeable visual incoherence, triggering artifacts such as aliasing or inconsistent sharpness under different viewpoints.
To address these issues, we propose PATexGS, a perceptual-adaptive texture scheduling framework designed to improve visual coherence for Textured Gaussians. Specifically, we introduce an entropy-guided texture allocation strategy that dynamically adjusts texture resolution based on each Gaussian’s spatial gradient and rendering contribution, constantly preserving details while being memory efficiency. Furthermore, we incorporate a mipmap-inspired hierarchical scheduling mechanism that adaptively schedule texture levels according to view-dependent projection scale, effectively suppressing aliasing and further enhancing perceptual consistency.
Extensive experiments on diverse real-world scenes demonstrate that PATexGS significantly improves visual coherence while maintaining high rendering quality, outperforming existing TexturedGS variants in both perceptual fidelity and storage efficiency.

PATexGS: Perceptual-Adaptive Texture Scheduling for Visual Coherence in Textured Gaussian Splatting

The massive scale of data and computation required for training Multimodal Large Language Models (MLLMs) has fueled the rise of Fine-Tuning as a Service (FTaaS), enabling users to rapidly customize models for diverse real-world tasks. While FTaaS democratizes access to advanced multimodal intelligence, it also introduces serious security concerns, particularly backdoor attacks. In this work, we systematically analyze backdoor vulnerabilities in MLLMs under the FTaaS paradigm, revealing two key phenomena: (1) markedly reduced sensitivity to textual variations when a visual trigger is present, and (2) abnormally stable model confidence even under strong semantic perturbations. Building on these insights, we propose **Trap on Text (ToT)**, a novel inference-time backdoor detection framework. ToT applies controlled semantic perturbations to textual prompts and jointly analyzes the **semantic consistency** and **confidence drift** of the model’s responses, enabling robust detection of backdoor activations without requiring model parameters, architectures or clean reference data. Extensive experiments across architectures and datasets show that ToT achieves strong attack mitigation and preserves clean accuracy, offering a practical solution for safeguarding FTaaS workflows.

Probing Semantic Insensitivity for Inference-Time Backdoor Defense in Multimodal Large Language Model

Graph neural networks (GNNs) can effectively model structural information of graphs, making them widely used in knowledge graph (KG) reasoning. However, existing studies on the expressive power of GNNs mainly focuses on simple single-relation graphs, and there is still insufficient discussion on the power of GNN to express logical rules in KGs. How to enhance the logical expressive power of GNNs is still a key issue. Motivated by this, we propose Path-Neighbor enhanced GNN (PN-GNN), a method to enhance the logical expressive power of GNN by aggregating node-neighbor embeddings on the reasoning path. First, we analyze the logical expressive power of existing GNN-based methods and point out the shortcomings of the expressive power of these methods. 
Then, we theoretically investigate the logical expressive power of PN-GNN, showing that it not only has strictly stronger expressive power than C-GNN but also that its $(k+1)$-hop logical expressiveness is strictly superior to that of $k$-hop. Finally, we evaluate the logical expressive power of PN-GNN on six synthetic datasets and two real-world datasets. Both theoretical analysis and extensive experiments confirm that PN-GNN enhances the expressive power of logical rules without compromising generalization, as evidenced by its competitive performance in KG reasoning tasks.

Enhancing Logical Expressiveness in Graph Neural Networks via Path-Neighbor Aggregation

Scene text image super-resolution aims to enhance text legibility by recovering high-resolution text images from low-resolution inputs. However, maintaining fine details like text strokes, edges, and textual accuracy remains challenging, especially in low-light and high-speed motion scenarios, where degradation is more severe. Event cameras, with their high temporal resolution and ability to capture intensity changes, offer a promising solution to restore lost fine details and mitigate degradation in these challenging conditions. In this paper, we propose EvTSR, the first framework that integrates Event data for scene Text image Super-Resolution. The core of EvTSR is the dual stream frequency boost (DSFB) mechanism, which separates image features into high- and low-frequency components. High-frequency details like edges and strokes are enhanced using event data via the event-guided high-frequency (EGH) module, while low-frequency components, responsible for global structure, are refined using the Text-Guided Low-frequency (TGL) module with a pre-trained text recognizer, ensuring textual coherence. To further improve cross-modal integration, we introduce the Cross-Modal Fusion (CMF) module, which effectively aligns event and image features, enabling robust information fusion. Extensive experiments show EvTSR achieves superior performance over existing methods.

Event-Guided Scene Text Image Super-Resolution

Cross-subject EEG decoding remains a fundamental challenge due to substantial inter-subject variability in brain activity, which hinders the development of subject-independent EEG models. Despite progress in extracting cross-subject invariant features, existing studies neglect the shared neural responses that arise under similar cognitive or emotional states across individuals, limiting their ability to learn generalized and consistent EEG representations. To address the challenges, we propose State Mamba, a novel spatiotemporal EEG state-space model that explicitly models and aligns neural responses and their spatiotemporal state transitions to learn consistent and generalizable representations across subjects. Innovatively, State Mamba theoretically formulates a multi-channel Mamba architecture that jointly models spatial and temporal brain state transitions, supporting principled analysis of neural responses. To enhance spatiotemporal feature coupling, we introduce the LGANN module, which adopts global-local attention to integrate long- and short-term brain activity into a compact EEG representation. Furthermore, we design two self-supervised pretext tasks to extract consistent neural patterns across subjects: (1) representation alignment to align EEG representation, and (2) pattern alignment to align their transition rules under identical conditions, jointly promoting subject-invariant EEG representations. Extensive experiments on three benchmark datasets, FACED, DEAP, and ISRUC, demonstrate the superior performance of State Mamba in cross-subject emotion and sleep recognition tasks, validating its robust generalization capability.

State Mamba: Spatiotemporal EEG State-Space Model with Dynamic Brain Alignment for Cross-Subject Representation

Advances in Multimodal Large Language Models have significantly enhanced Graphical User Interface (GUI) automation. Equipping GUI agents with reliable episodic reasoning capabilities is essential for bridging the gap between users’ concise task descriptions and the complexities of real-world execution. Current methods integrate Reinforcement Learning (RL) with System-2 Chain-of-Thought, yielding notable gains in reasoning enhancement. For long-horizon GUI tasks, historical interactions connect each screen to the goal-oriented episode chain, and effectively leveraging these clues is crucial for the current decision. However, existing native GUI agents exhibit weak short-term memory in their explicit reasoning, interpreting the chained interactions as discrete screen understanding, i.e., unawareness of the historical interactions within the episode. This history-agnostic reasoning challenges their performance in GUI automation. To alleviate this weakness, we propose a History-Aware Reasoning (HAR) framework, which encourages an agent to reflect on its own errors and acquire episodic reasoning knowledge from them via tailored strategies that enhance short-term memory in long-horizon interaction. The framework mainly comprises constructing a reflective learning scenario, synthesizing tailored correction guidelines, and designing a hybrid RL reward function. Using the HAR framework, we develop a native end-to-end model, HAR-GUI-3B, which alters the inherent reasoning mode from history-agnostic to history-aware, equipping the GUI agent with stable short-term memory and reliable perception of screen details. Comprehensive evaluations across a range of GUI-related benchmarks demonstrate the effectiveness and generalization of our method.

History-Aware Reasoning for GUI Agents

Vision-language-action models have emerged as a crucial paradigm in robotic manipulation. However, existing VLA models exhibit notable limitations in handling ambiguous language instructions and unknown environmental states. Furthermore, their perception is largely constrained to static two-dimensional observations, lacking the capability to model three-dimensional interactions between the robot and its environment. To address these challenges, this paper proposes GraphCoT-VLA, an efficient end-to-end model. To enhance the model's ability to interpret ambiguous instructions and improve task planning, we design a structured Chain-of-Thought reasoning module that integrates high-level task understanding and planning, failed task feedback, and low-level imaginative reasoning about future object positions and robot actions. Additionally, we construct a real-time updatable 3D Pose-Object graph, which captures the spatial configuration of robot joints and the topological relationships between objects in 3D space, enabling the model to better understand and manipulate their interactions. We further integrates a dropout hybrid reasoning strategy to achieve efficient control outputs. Experimental results across multiple real-world robotic tasks demonstrate that GraphCoT-VLA significantly outperforms existing methods in terms of task success rate and response speed, exhibiting strong generalization and robustness in open environments and under uncertain instructions.

Downloads

Next from AAAI 2026

Revealing the Invisible: Latent Structure Modeling for Semantically Consistent Cloud Removal

Similar lecture

ViTCoP: Accelerating Large Vision-Language Models via Visual and Textual Semantic Collaborative Pruning

.css-70qvj9{display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;-webkit-align-items:center;-webkit-box-align:center;-ms-flex-align:center;align-items:center;}Downloads

Next from AAAI 2026

Revealing the Invisible: Latent Structure Modeling for Semantically Consistent Cloud Removal

Similar lecture

ViTCoP: Accelerating Large Vision-Language Models via Visual and Textual Semantic Collaborative Pruning

Downloads