Singapore

Although current large audio language models (LALMs) extend text large language models (LLMs) with generic acoustic understanding abilities, they usually suffer from instruction sensitivity, where different instructions of the same intention can yield drastically different outcomes. In this work, we propose AHAMask, where we simply mask some of the attention heads in the decoder-only LLM backbone of LALMs, to trigger specific acoustic task functionalities without instructions. These masks are efficiently obtained by training on an LALM, with the number of trainable parameters equal to the attention head count in its LLM backbone. We show by experiments that applying such selective attention head masks achieves comparable or even better performance than using instructions, either on single or composite tasks. Besides achieving reliable acoustic task specification for LALMs, this also reveals that LALMs exhibit certain ``functional pathways&#39;&#39; in their attention heads. We will open-source the codebase after the anonymity period.

AAAI 2026

AHAMask: Reliable Task Specification for Large Audio Language Models Without Instructions

functional partitioning

attention head mask

prompt sensitivity

### Welcome to The 63rd Annual Meeting of the Association for Computational Linguistics (ACL 2025)

Message from the General Chair: 
*It is my great pleasure and honor to welcome you to the 63rd Annual Meeting of the Association for Computational Linguistics (ACL 2025), held in beautiful Vienna, Austria, from July 27 to August 1, 2025. ACL2025continues our field’s tradition of excellence in scholarship, innovation, and inclusivity, and I am deeply grateful to the many volunteers who have worked tirelessly to bring this event to life.* 
[Read more](https://drive.google.com/file/d/1GI_hvOpjswAuYdUTromfeDiPpCcqidwg/view?usp=sharing)

To access this event page, you need to log in with the **email address you registered with**. Access credentials will be sent to your email from Underline - subject line "Welcome to ACL 2025". 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 need to log in with the **email address you registered with**. 

ACL 2025

Welcome to The 63rd Annual Meeting of the Association for Computational Linguistics (ACL 2025)

Although current large audio language models (LALMs) extend text large language models (LLMs) with generic acoustic understanding abilities, they usually suffer from instruction sensitivity, where different instructions of the same intention can yield drastically different outcomes. In this work, we propose AHAMask, where we simply mask some of the attention heads in the decoder-only LLM backbone of LALMs, to trigger specific acoustic task functionalities without instructions. These masks are efficiently obtained by training on an LALM, with the number of trainable parameters equal to the attention head count in its LLM backbone. We show by experiments that applying such selective attention head masks achieves comparable or even better performance than using instructions, either on single or composite tasks. Besides achieving reliable acoustic task specification for LALMs, this also reveals that LALMs exhibit certain ``functional pathways'' in their attention heads. We will open-source the codebase after the anonymity period.

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.

To access this event page, you need to log in with the **email address you registered with**. 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.

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.

Time Series forecasting (TSF) in the modern era faces significant computational and storage cost challenges due to the massive scale of real-world data. Dataset Distillation (DD), a paradigm that synthesizes a small, compact dataset to achieve training performance comparable to that of the original dataset, has emerged as a promising solution. However, conventional DD methods are not tailored for time series and suffer from architectural overfitting and limited scalability. To address these issues, we propose Harmonic Dataset Distillation for Time Series Forecasting (HDT). HDT decomposes the time series into its sinusoidal basis through the FFT and aligns the core periodic structure by Harmonic Matching. Since this process operates in the frequency domain, all updates during distillation are applied globally without disrupting temporal dependencies of time series. Extensive experiments demonstrate that HDT achieves strong cross-architecture generalization and scalability, validating its practicality for large-scale, real-world applications.

Harmonic Dataset Distillation for Time Series Forecasting

Understanding when a pre-trained model generalizes well to a new task remains a key challenge in transfer learning. Classical theories bound target risk using divergences such as total variation, MMD, or Wasserstein distance, yet tasks with similar divergences often show very different transfer performance. We propose a structural framework that explains transferability through two factors: the Feature Overlap Rate (FOR), measuring how much target representation lies in the source-induced subspace, and the Effective Task Complexity (ETC), quantifying the entropy of latent subtasks. We derive a PAC-Bayesian bound where target risk depends on FOR and ETC, and show that larger models attenuate their negative effects. Experiments on six GLUE transfer pairs estimate FOR and ETC from encoder representations and compare them to classical divergences. Results show that FOR and ETC together explain over 80% of transfer risk variance, while divergences fail to do so. Our findings provide a geometry-aware perspective for diagnosing and guiding transfer learning.

Unified Structural Factors for Transfer Learning Generalization with PAC-Bayesian Guarantees

Aerial multi-modal visual streams registration and fusion can generate more comprehensive scene information representations for UAVs' cross-modal perception. However, current challenges lie primarily in the essential difficulty of joint spatiotemporal representation learning from dynamic background and moving targets, and a critical shortage exists in large-scale, well-annotated multi-modal visual streams benchmark for UAV platforms. In this paper, we propose AerialFusion, a co-motion-driven unified UAVs visual streams registration and fusion that fully mines modality-invariant common features based on motion-aware, enabling spatiotemporally coherent registration and fusion. Specifically, 1) a Skewed Motion Distribution Field Co-Motion-Driven Image Registration, 2) a Co-Motion Generative Fusion, 3) a Streams-based Unified Learning. Furthermore, we introduce EUM3D, a registration and fusion benchmark for UAVs cross-modal perception. This benchmark contains 60 synchronized visible-infrared visual streams, or 122k spatially and temporally aligned pairs, most of which were taken at low-light scenes. And EUM3D provides pixel-level alignment guarantees via perspective-transform ground-truth. Extensive experiments reveal that AerialFusion surpasses current focus on image and static background fusion methods in aerial sequence scenarios, addressing spatiotemporal mismatches while suppressing cross-modal interference.

AerialFusion: Co-Motion-Driven Unified Registration and Fusion on Multi-modal Data Streams from Aerial View

Large Language Models (LLMs) hold significant potential for enhancing healthcare applications, yet their deployment is hindered by high computational and memory demands. Model compression techniques, such as quantization and sparsification, offer solutions to reduce these demands, but their impact on medical LLMs remains underexplored. We introduce CMedBench, the first comprehensive benchmark for evaluating compressed LLMs in medical contexts. LLMCMedBench assesses five core dimensions: Medical Knowledge Ability, Medical Application Ability, Trustworthiness Maintenance, Compression Cross Combination, and Computational Efficiency. Through extensive empirical studies, we analyze the trade-offs between model efficiency and clinical performance across diverse models, datasets, and compression strategies. Our findings highlight critical limitations in current evaluation practices and provide a robust framework for aligning compression strategies with medical requirements. CMedBench serves as a vital resource for researchers and practitioners, guiding the development of efficient, trustworthy, and clinically effective LLMs for healthcare applications.

CMedBench: A Comprehensive Benchmark for Efficient Medical Large Language Models

Large Language Models (LLMs) have demonstrated strong performance in various NLP tasks but remain limited in emotional intelligence (EI). Benchmarks such as EmoBench attribute this gap to deficiencies in cognitively demanding tasks that require inferring others’ latent mental states, intentions, and emotions in nuanced social contexts. To address this, we propose MACRo, a Multi-Agent Cognitive Reasoning framework that generates a structured Cognitive Chain of Thought comprising Situation, Clue, Thought, Action, and Emotion. Each component is generated by a specialized agent, enabling modular, interpretable multi-step reasoning. To ensure coherence and mitigate hallucinations, a coordinator agent verifies outputs, and a consensus game mechanism enforces alignment across reasoning steps. Extensive Experiments on EmoBench show that MACRo significantly enhances both emotional understanding and application across LLMs. Further evaluations confirm its generalizability to real-world social applications such as emotional support conversations.

Consensus-Driven Multi-Agent Cognitive Reasoning for Enhancing the Emotional Intelligence of Large Language Models

Counterfactual decision-making in the face of uncertainty involves selecting the optimal action from several alternatives using causal reasoning. Decision-makers often rank expected potential outcomes (or their corresponding utility and desirability) to compare the preferences of candidate actions. In this paper, we study new counterfactual decision-making rules by introducing two new metrics: the probabilities of potential outcome ranking (PoR) and the probability of achieving the best potential outcome (PoB). PoR reveals the most probable ranking of potential outcomes for an individual, and PoB indicates the action most likely to yield the top-ranked outcome for an individual. We then establish identification theorems and derive bounds for these metrics, and present estimation methods. Finally, we perform numerical experiments to illustrate the finite-sample properties of the estimators and demonstrate their application to a real-world dataset.

Potential Outcome Rankings for Counterfactual Decision Making

Post-Training Quantization enables efficient Vision Transformer (ViTs) deployment with a small calibration data, and its prevalent use of uniform quantization harnesses AI accelerator matrix cores for high-speed inference. However, the application of uniform quantization is fundamentally challenged by the extreme non-uniformity of activation distributions.Specifically, the power-law nature of post-Softmax attention scores and the significant inter-channel variance in post-GELU activations create a dilemma for conventional quantization, as it struggles to preserve critical high-magnitude values without sacrificing overall precision. To resolve this core conflict, we introduce UQ-ViT (Uniform Quantization for Vision Transformers), a novel uniform quantization framework designed to reconcile high precision with hardware efficiency. Central to UQ-ViT are two operators: Dynamic Elimination of Maximum (DeMax) and Normalization Quantization (NormQuant). DeMax is a quantization operator for post-Softmax attention scores that utilizes uniform quantization. It dynamically eliminates and preserves dominant values, effectively mitigating quantization loss from the extreme values in the power-law distribution. NormQuant utilizes a per-channel quantization strategy during quantization and reverts to a per-tensor format for dequantization, achieving both high accuracy and computational efficiency. Crucially, it is applicable to any linear layer, enabling effective quantization of post-GELU activations in ViTs. Through extensive experiments on various ViTs and vision tasks, including image classification, object detection, and instance segmentation, we demonstrate that our proposed approach outperforms existing methods, achieving superior accuracy while ensuring hardware friendliness.

UQ-ViT: Harmonizing Extreme Activations with Hardware-Friendly Uniform Quantization in Vision Transformers

The proliferation of the tampered images on social media can pose serious societal risks, influencing public opinion and causing panic. Image Manipulation Localization technique has advanced to address this, but some methods focus on microscopic traces, overlooking macroscopic semantics that deceive viewers. To address this problem, we propose a novel Image Manipulation Localization framework called Collaborative Transformers (Co-Transformers), designed to fully explore and utilize the collaborative information between macroscopic semantics and microscopic traces. This framework is based on two Vision Transformer variants. The first variant captures the semantic logic of the image. The second variant delves into microscopic tampering traces. By dynamically fusing these two complementary features, the framework enables interaction between macroscopic semantic inconsistencies and microscopic abnormal traces, effectively coordinating their relationship in the latent space. Furthermore, we introduce a new Multi-Level Forensic Attention (MLF-Attention) mechanism to enhance the model's ability to extract various tampered traces, this mechanism can be integrated into our framework. Compared with existing methods, our proposed framework achieves state-of-the-art results in localization accuracy and shows good robustness against various attacks.

Collaborative Transformers with Multi-Level Forensic Attention for Image Manipulation Localization

Current low-light image restoration methods suffer from severe efficiency bottlenecks, primarily stemming from: (1) computational burden and error correction costs associated with reliance on external priors (manual or cross-modal); (2) redundant operations in complex multi-stage enhancement pipelines; and (3) indiscriminate processing across frequency components in frequency-domain methods, leading to excessive global computational demands. To address these challenges, we propose an Efficient Self-Mining Prior-Guided Joint Frequency Enhancement Network (SPJFNet). Specifically, we first introduce a Self-Mining Guidance Module (SMGM) that generates lightweight endogenous guidance directly from the network, eliminating dependence on external priors and thereby bypassing error correction overhead while improving inference speed. Second, through meticulous analysis of different frequency domain characteristics, we reconstruct and compress multi-level operation chains into a single efficient operation via lossless wavelet decomposition and joint Fourier-based advantageous frequency enhancement, significantly reducing parameters. Building upon this foundation, we propose a Dual-Frequency Guidance Framework (DFGF) that strategically deploys specialized high/low frequency branches (wavelet-domain high-frequency enhancement and Fourier-domain low-frequency restoration), decoupling frequency processing to substantially reduce computational complexity. Rigorous evaluation across multiple benchmarks demonstrates that SPJFNet not only surpasses state-of-the-art performance but also achieves significant efficiency improvements, substantially reducing model complexity and computational overhead.

SPJFNet: Self-Mining Prior-Guided Joint Frequency Enhancement for Ultra-Efficient Dark Image Restoration

Recent advancements in Large Language Models (LLMs) have increasingly demonstrated their potential for event reasoning. However, current LLMs struggle with this task due to their limited capacity to explicitly model the structured semantics of events, resulting in insufficient schema knowledge and low reasoning performance. To address these challenges, we propose SGER, a Schema-Guided Event Reasoning framework. It constructs a systematic solution by decomposing complex event reasoning tasks into three interrelated subtasks: schema extraction, schema prediction, and event reasoning. In the schema extraction stage, the model maps event descriptions with diverse surface forms to potential semantic structure representations, achieving an abstract transformation from instances to schemas. The schema prediction stage captures the potential associations between historical event schemas to make forward-looking inferences about possible future event schemas. In the event reasoning stage, we integrate historical events and predicted schemas into prompts to guide LLMs in generating specific, contextually consistent predicted events. Experimental evaluations demonstrate that our framework significantly improves event reasoning performance of LLMs.

Downloads

Next from AAAI 2026

Harmonic Dataset Distillation for Time Series Forecasting

Similar lecture

Benchmarking Open-ended Audio Dialogue Understanding for Large Audio-Language Models