Singapore

Recent advances demonstrate that reinforcement learning with verifiable rewards (RLVR) significantly enhances the reasoning capabilities of large language models (LLMs). However, standard RLVR faces challenges with reward sparsity, where zero rewards from consistently incorrect candidate answers provide no learning signal, particularly in challenging tasks. To address this, we propose, we propose \textbf{M}ulti-\textbf{E}xpert \textbf{M}utual \textbf{L}earning GRPO (MEML-GRPO), an innovative framework that utilizes diverse expert prompts as system prompts to generate a broader range of responses, substantially increasing the likelihood of identifying correct solutions. Additionally, we introduce an inter-expert mutual learning mechanism that facilitates knowledge sharing and transfer among experts, further boosting the model’s performance through RLVR. Extensive experiments across multiple reasoning benchmarks show that MEML-GRPO delivers significant improvements, achieving an average performance gain of 4.89\% with Qwen and 11.33\% with Llama, effectively overcoming the core limitations of traditional RLVR methods.

AAAI 2026

MEML-GRPO: Heterogeneous Multi-Expert Mutual Learning for RLVR Advancement

rlvr

generation

poster

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.

Machine unlearning is a newly popularized technique for removing specific training data from a trained model, enabling it to comply with data deletion requests. While it protects the rights of users requesting unlearning, it also introduces new privacy risks. Prior works have primarily focused on the privacy of data that has been unlearned, while the risks to retained data remain largely unexplored. 
To address this gap, we focus on the privacy risks of retained data and, for the first time, reveal the vulnerabilities introduced by machine unlearning under the \textbf{dual-view} setting, where an adversary can query both the original and the unlearned models. From an information-theoretic perspective, we introduce the concept of {privacy knowledge gain} and demonstrate that the dual-view setting allows adversaries to obtain more information than querying either model alone, thereby amplifying privacy leakage. To effectively demonstrate this threat, we propose DVIA, a \textbf{D}ual-\textbf{V}iew \textbf{I}nference \textbf{A}ttack, which extracts membership information on retained data using black-box queries to both models. DVIA eliminates the need to train an attack model and employs a lightweight likelihood ratio inference module for efficient inference. Experiments across different datasets and model architectures validate the effectiveness of DVIA and highlight the privacy risks inherent in the dual-view setting.

Dual-View Inference Attack: Machine Unlearning Amplifies Privacy Exposure

Image clustering is a fundamental task in unsupervised visual learning. While recent self-supervised methods have explored various pretext tasks to generate supervision signals for clustering, they typically depend exclusively on raw images, resulting in insufficient supervision signals that are inherently constrained by limited visual semantics. In this paper, we propose a novel Semantic-Augmented image Clustering (SAC) method, which transcends the inherent limitations of purely visual representations through the integration of external knowledge. Specifically, SAC utilizes Vision-Language pre-trained Models (VLMs) to flexibly generate textual descriptions for each image, providing external semantic cues to supplement the visual information. By integrating both visual and textual information, SAC achieves image clustering through a multi-modal learning framework. To mitigate the negative impact of inaccurate textual information, SAC designs an uncertainty-driven adaptive weighting mechanism that explores both intra-modal and inter-modal neighborhood structures, and incorporates the adaptive weights into intra-modal and inter-modal contrastive learning, which improves the robustness against noisy image-text correspondences. Experiments on several popular datasets demonstrate the superiority of SAC compared to state-of-the-art methods.

Semantic-Augmented Image Clustering via Adaptive Multi-Modal Collaboration

Micro-video label prediction plays a pivotal role on contemporary video-sharing platforms, such as Kwai and Tiktok. The emergence of video content lacking labels presents a formidable challenge for conventional user interest prediction methods. This paper addresses the challenge of micro-video label prediction, particularly for unseen videos, by proposing a zero-shot method called Class Semantic Relation Learning (CSRL). Unlike traditional user interest prediction models, CSRL leverages the pre-trained Large Language Model (LLM) to enhance prediction accuracy for unlabeled videos. The novelty of CSRL lies in its integration of three key components: a raw feature autoencoder, LLM-enhanced features, and a decomposed graph network. The decomposed graph network is specifically designed to disentangle the relationships between labeled and unlabeled videos, offering a significant improvement over previous methods. By fusing hidden topics with LLM-enhanced text, CSRL effectively handles sparse video features. Experiments on large-scale datasets from the Kwai platform show that CSRL achieves state-of-the-art results, with up to 44.64\% improvement in Hit Ratio (HR), highlighting its superiority over existing zero-shot recommendation models in predicting user interests within the user-video network.

Zero-shot Recommendation: Towards Class Semantic Relation Learning for Inferring Labels of Unseen Micro-videos

Existing open-source datasets for arbitrary-instruction image editing remain suboptimal, while a plug-and-play editing module compatible with community-prevalent generative models is notably absent. In this paper, we first introduce the X2Edit Dataset, a comprehensive dataset covering 14 diverse editing tasks, including subject-driven generation. We utilize the industry-leading unified image generation models and expert models to construct the data. Meanwhile, we design reasonable editing instructions with the VLM and implement various scoring mechanisms to filter the data. As a result, we construct 3.7 million high-quality data with balanced categories. Second, to better integrate seamlessly with community image generation models, we design task-aware MoE-LoRA training based on FLUX.1, with only 8% of the parameters of the full model. To further improve the final performance, we utilize the internal representations of the diffusion model and define positive/negative samples based on image editing types to introduce contrastive learning. Extensive experiments demonstrate that the model's editing performance is competitive among many excellent models. Additionally, the constructed dataset exhibits substantial advantages over existing open-source datasets.

X2Edit: Revisiting Arbitrary-Instruction Image Editing Through Self-Constructed Data and Task-Aware Representation Learning

Singular Value Decomposition (SVD) has recently seen a surge of interest as a simple yet powerful tool for large language models (LLMs) compression, with a growing number of works demonstrating 20 - 80\% parameter reductions at minimal accuracy loss. Previous SVD-based approaches have focused primarily on reducing the memory footprint of model weights, largely overlooking the additional activation memory overhead incurred during inference when applying truncated factors via standard dense CUDA kernels. Our experiments demonstrate that this activation overhead, scaling with sequence length and hidden dimension, prevents current SVD compression techniques from achieving any reduction in peak inference memory, thereby limiting their viability for real-world, on-device deployments. 

We introduce \textbf{FlashSVD}, a novel, end-to-end rank-aware streaming inference framework specifically designed for SVD-compressed large language models. FlashSVD can be seamlessly integrated with any model that employs SVD-based methods for parameter reduction. By fusing low-rank projection kernels directly into both the self-attention and feed-forward network (FFN) pipelines, FlashSVD avoid materializing full‐size activation buffers. Instead, small tiles of the truncated factors are loaded into on-chip SRAM, multiplied and reduced on the fly, and immediately evicted, preserving high GPU occupancy and adding no extra latency. On standard encoder benchmarks (e.g., BERT-Base), FlashSVD cuts peak activation memory by up to 70.2\% and intermediate transient memory by 75\%, all while incur no accuracy loss with upstreaming compression methods, offering a practical path toward memory-constrained deployment of low-rank LLMs.

FlashSVD: Memory-Efficient Inference with Streaming for Low-Rank Models

Generating sketches with specific patterns as expected, i.e., manipulating sketches in a controllable way, is a popular task. Recent studies control sketch features at stroke-level by editing values of stroke embeddings as conditions. However, in order to provide generator a global view about what a sketch is going to be drawn, all these edited conditions should be collected and fed into generator simultaneously before generation starts, i.e., no further manipulation is allowed during sketch generating process. In order to realize sketch drawing manipulation more flexibly, we propose a hierarchical auto-regressive sketch generating process. Instead of generating an entire sketch at once, each stroke in a sketch is generated in a three-staged hierarchy: 1) predicting a stroke embedding to represent which stroke is going to be drawn, and 2) anchoring the predicted stroke on the canvas, and 3) translating the embedding to a sequence of drawing actions to form the full sketch. Moreover, the stroke prediction, anchoring and translation are proceeded auto-regressively, i.e., both the recently generated strokes and their positions are considered to predict the current one, guiding model to produce an appropriate stroke at a suitable position to benefit the full sketch generation. It is flexible to manipulate stroke-level sketch drawing at any time during generation by adjusting the exposed editable stroke embeddings.

Generating Sketches in a Hierarchical Auto-Regressive Process for Flexible Sketch Drawing Manipulation at Stroke-Level

In this work, we introduce a novel high-fidelity full-head 3D avatar generation method from a single image, regardless of perspective, style, expression, or accessories. Prior works often fail to preserve consistent head geometry and facial details, primarily due to their limited capacity in modeling fine-grained facial textures and maintaining identity information. To address these challenges, we construct a new high-quality dataset containing 227 sequences of digital human portraits captured from 96 different perspectives, totalling 21,792 frames, featuring high-quality facial texture details. To further improve performance, we propose a novel multi-view diffusion named ID-TS diffusion model, which integrate identity and expression information into the two-stage multi-view diffusion process. The low-resolution stage ensures structural consistency of heads across multiple views, while the high-resolution stage preserves facial detail fidelity and coherence. Finally, we propose an enhanced feed-forward Gaussian avatar reconstruction method that optimizes the network on multi-view images of each single subject, significantly improving 3D facial texture details. Extensive experiments show that our method demonstrates robust performance across challenging scenarios, while showcasing broad applicability across numerous downstream tasks.

High-Quality Full-Head 3D Avatar Generation from Any Single Portrait Image

Semi-supervised partial label learning (SSPLL) aims to improve the generalization performance of partial label (PL) classifiers by effectively leveraging unlabeled data. Nevertheless, the inherent ambiguity in supervision, where the ground-truth label of a PL example is hidden within a set of candidate labels, poses significant challenges. The presence of false positive labels potentially misleads model's judgment, resulting in pronounced confirmation bias. To address these issues, we propose a novel approach named CODUAL, which jointly learns a pair of dual representations for each instance: the predictive class distribution and the low-dimensional embedding. The dual representations interact and progress collaboratively during training. On one hand, in the embedding space the class prototypes are derived via solving a tailored empirical distance minimization problem and employed to smooth the pseudo-targets of unlabeled instances. On the other hand, the refined class distributions regularize the embedding space via encouraging instances with similar pseudo-targets to exhibit similar embeddings. Through an in-depth analysis, we provide-to the best of our knowledge-the first theoretical explanation of how collaborative dual representations facilitate more effective use of unlabeled data for disambiguation. Extensive experiments over benchmark datasets validate the superiority of our proposed approach.

Collaborative Dual Representations for Semi-Supervised Partial Label Learning

Continuous learning constitutes a fundamental capability of artificial intelligence systems, enabling them to incrementally assimilate novel information without succumbing to catastrophic forgetting. Recent research has leveraged Pre-Trained Models (PTMs) to enhance continual learning efficacy. Nevertheless, prevailing methodologies typically depend on a singular pre-trained backbone and freeze all pre-trained parameters to mitigate network forgetting, thereby constraining adaptability to emerging tasks. In this study, we introduce an innovative PTM-based framework featuring a Dual-Representation Backbone Architecture (DRBA), which integrates both invariant and evolved representation networks to concurrently capture static and dynamic features. Building upon DRBA, we propose an Adaptive and Expandable Mixture Model (AEMM) that incrementally incorporates new expert modules with minimal parameter overhead to accommodate the learning of each novel task. To further augment adaptability, we develop a Dynamic Adaptive Representation Fusion Mechanism (DARFM) that processes outputs from both representation networks and autonomously generates data-driven adaptive weights, optimizing the contribution of each representation. This mechanism yields an adaptive, semantically enriched composite representation, thereby maximizing positive knowledge transfer. Additionally, we propose a Dynamic Knowledge Calibration Mechanism (DKCM), comprising prediction and representation calibration processes, to ensure consistency in both predictions and feature representations. This approach achieves a balance between stability and plasticity, even when learning complex datasets. Empirical evaluations substantiate that the proposed approach attains state-of-the-art performance.

Learning Adaptive and Expandable Mixture Model for Continual Learning

3D human avatar animation aims at transforming a human avatar from an arbitrary initial pose to a specified target pose using deformation algorithms. Existing approaches typically divide this task into two stages: canonical template construction and target pose deformation. However, current template construction methods demand extensive skeletal rigging and often produce artifacts in contact regions. Moreover, target pose deformation suffers from structural distortions caused by Linear Blend Skinning (LBS), which significantly undermines animation realism. To address these problems, we propose a unified learning-based framework to address both challenges in two phases. For the former phase, to overcome the inefficiencies and artifacts during template construction, we leverage a U-Net architecture that decouples texture and pose information in a feed-forward process, enabling fast generation of a human template. For the latter phase, we propose a data-driven refinement technique that enhances structural integrity. Extensive experiments show that our model delivers consistent performance across diverse poses with an optimal balance between efficiency and quality, surpassing state-of-the-art (SOTA) methods.

Content not yet available

Next from AAAI 2026

Dual-View Inference Attack: Machine Unlearning Amplifies Privacy Exposure

Stay up to date with the latest Underline news!

PRESENTATIONS

CONFERENCES

COMPANY

RESOURCES