Singapore

Self-evaluation, a model&#39;s ability to assess the correctness of its own output, is crucial for Large Multimodal Models (LMMs) to achieve self-improvement in multi-turn conversations, yet largely absent in foundation models. Recent work has employed reinforcement learning (RL) to enhance self-evaluation; however, its fixed reward mechanism suffers from reward hacking when optimizing multiple training objectives, leading to model collapse. In this paper we propose AdaPO, an online reinforcement learning framework capable of adaptively adjusting training objective in real time according to the current training state for each task. Specifically, to mitigate reward hacking , AdaPO introduces an Adaptive Reward Model (ARM) and a Reward Aware Dynamic KL Regularization mechanism. ARM assesses the task&#39;s training state from the distribution of model generated multi-turn trajectories&#39; performance. Reward Aware Dynamic KL replaces a fixed penalty with dynamic coefficients which is modulated by the reward gap between different multi-turn situations. 
Notably, our method automatically and smoothly adjusts its learning focus based on sub-tasks&#39; training progress without manual intervention.
Extensive experiments over 8 benchmarks and various models show that our method significantly enhances both direct reasoning and self-evaluation capability. We will release our code to contribute to the community.

AAAI 2026

A Rolling Stone Gathers No Moss: Adaptive Policy Optimization for Stable Self-Evaluation in Large Multimodal Models

language grounding & multi-modal nlp

(large) language models

natural language processing (nlp)

It is my great pleasure and honor to welcome you to the 64th Annual Meeting of the Association for Computational Linguistics (ACL 2026), held in San Diego, California, USA, July 2–7, 2026. ACL 2026 continues our fields 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.

It is exciting to see the many developments in the field reflected at our conference, such as the advancement of large language models (born out of our community) into many application directions. This is not without challenges, partly due to unsolved problems, and partly due to misuse of the technology. All this is reflected well at ACL 2026. The conference also reflects that the field is not only about advancing technology, but also the science of fundamentally understanding language and how it is use by many different speaker communities. 
[Full Message from General Chair Philipp Koehn](https://drive.google.com/file/d/1QiJFOmMjvDgsDo3i1XzlQnImxe3RU_Cy/view?usp=sharing) 
<html><button style="display: inline-flex; align-items: center; justify-content: center; white-space: nowrap; border-radius: 9999px; font-weight: bold; background: #7c3aed; color: white; font-family: 'Space Grotesk', sans-serif; height: 40px; font-size: 16px; padding: 0 20px; border: none; cursor: pointer" onclick="window.open('https://drive.google.com/file/d/1QiJFOmMjvDgsDo3i1XzlQnImxe3RU_Cy/view?usp=sharing')">Conference Handbook</button></html>

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 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://2026.aclweb.org/registration/) first.

**Online Registration Form**: https://acl.swoogo.com/acl2026

Registration Required

ACL 2026

The 64th Annual Meeting of the Association for Computational Linguistics (ACL 2026) will take place in San Diego, California, from July 2 - 7, 2026.

Self-evaluation, a model's ability to assess the correctness of its own output, is crucial for Large Multimodal Models (LMMs) to achieve self-improvement in multi-turn conversations, yet largely absent in foundation models. Recent work has employed reinforcement learning (RL) to enhance self-evaluation; however, its fixed reward mechanism suffers from reward hacking when optimizing multiple training objectives, leading to model collapse. In this paper we propose AdaPO, an online reinforcement learning framework capable of adaptively adjusting training objective in real time according to the current training state for each task. Specifically, to mitigate reward hacking , AdaPO introduces an Adaptive Reward Model (ARM) and a Reward Aware Dynamic KL Regularization mechanism. ARM assesses the task's training state from the distribution of model generated multi-turn trajectories' performance. Reward Aware Dynamic KL replaces a fixed penalty with dynamic coefficients which is modulated by the reward gap between different multi-turn situations. 
Notably, our method automatically and smoothly adjusts its learning focus based on sub-tasks' training progress without manual intervention.
Extensive experiments over 8 benchmarks and various models show that our method significantly enhances both direct reasoning and self-evaluation capability. We will release our code to contribute to the community.

technical paper

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.

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 

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.

Generative recommendation (GR) is an emerging paradigm that tokenizes items into discrete tokens and learns to autoregressively generate the next tokens as predictions. While this token-generation paradigm is expected to surpass traditional transductive methods, potentially generating new items directly based on semantics, we empirically show that GR models predominantly generate items seen during training and struggle to recommend unseen items. In this paper, we propose SpecGR, a plug-and-play framework that enables GR models to recommend new items in an inductive setting. SpecGR uses a *drafter* model with inductive capability to propose candidate items, which may include both existing items and new items. The GR model then acts as a *verifier*, accepting or rejecting candidates while retaining its strong ranking capabilities. We further introduce the guided re-drafting technique to make the proposed candidates more aligned with the outputs of generative recommendation models, improving the verification efficiency. We consider two variants for drafting: (1) using an auxiliary drafter model for better flexibility, or (2) leveraging the GR model's own encoder for parameter-efficient self-drafting. Extensive experiments on three real-world datasets demonstrate that SpecGR exhibits both strong inductive recommendation ability and the best overall performance among the compared methods.

Inductive Generative Recommendation via Retrieval-based Speculation

Alerts generated by Security Operations Centers (SOCs) are often numerous and scattered, requiring significant effort from security analysts to manage, which severely slows response times.
While recent alert correlation graph methods can effectively reduce alert volume, these graphs are often too complex for analysts to understand.
As a result, analysts are increasingly seeking ways to automatically correlate alerts and generate concise, human-readable attack path summaries.
Recently, Large Language Models (LLMs) have demonstrated superior performance due to their advanced capabilities in knowledge reserve and reasoning.
In this work, we propose GARNET, a framework that uses LLMs for reasoning on alert correlation graphs.
GARNET addresses three key technical challenges: 1) modality alignment between alert graphs and logs; 2) semantic alignment between alert graphs and logs; 3) enabling LLMs reasoning along graph paths.
Specifically, we first project the embeddings of the graph and logs into the same vector space using contrastive learning. Then, we design self-supervised graph-log instructions to bridge the semantic gap between the graph and logs by training a novel LLM. Finally, GARNET uses a novel Graph-of-Thought (GoT)-based interaction reasoning approach to guide LLM reasoning along graph paths, ultimately generating structured, concise, and human-readable attack path summaries.
Experimental results across six attack scenarios show that GARNET reduces false positives by an average of 80\%, lowering the false positive rate to below 0.0037. It outperforms the latest approaches and provides more explainable attribution.

GARNET: GoT-Based Alert Reduction and Narrative Event Tracing

Incomplete multi-view clustering (IMVC) aims to group data into meaningful clusters when each sample is only partially observed across multiple views. Most existing methods either rely on imputation strategies that may introduce noise and distort the underlying data distribution, or adopt cross-view alignment techniques that focus on pairwise relationships, often resulting in suboptimal representations and unstable clustering performance. In this paper, we propose **G**eometry-**A**ware **V**ariational **I**nformation **M**aximization for Deep Incomplete Multi-view Clustering (GAVIM), a novel imputation-free variational framework that enables robust and coherent incomplete multi-view clustering. Specifically, GAVIM leverages mutual information maximization to preserve the high mutual information between the available multi-view data and the shared embedding. Morever, we explicitly retain local geometric consistency within each view-specific latent space under the guidance of an adaptive global supervision signal. Lastly, GAVIM aligns all views simultaneously using a Gramian representation alignment measure, ensuring coherent structure across modalities and promoting unified, semantically meaningful representations. Extensive experiments on five benchmark IMVC datasets with varying levels of view incompleteness demonstrate that GAVIM consistently outperforms state-of-the-art methods in clustering accuracy and representation quality.

Geometry-Aware Variational Information Maximization for Deep Incomplete Multi-view Clustering

Early childhood is a critical stage for cognitive development, involving core skills such as visual perception and reasoning. While multimodal large language models (MLLMs) have made rapid progress in various general-purpose tasks, their ability to support early education remains largely underexplored. Existing research on child-related AI largely centers on modeling language, emotion, or behavior, with limited focus on evaluating cognitive tasks relevant to early learning. To address this gap, we propose ChildBench, a multimodal benchmark designed to assess models on tasks inspired by early childhood cognitive development. It covers five key domains through ten tasks, including spatial reasoning, visual reasoning, visual discrimination, counting skills, and visual tracking. The benchmark includes 4,890 carefully constructed images and 5,346 manually annotated samples, ensuring both diversity and age-appropriate content. We evaluate a range of state-of-the-art (SoTA) open-source and closed-source MLLMs—including GPT-4o, Gemini, and Qwen2.5-VL—on ChildBench. Despite strong performance on other benchmarks, the best 7B-parameter model with LoRA tuning achieves only 52.01% accuracy, far below the 96% achieved by 5-year-old children. These results reveal critical limitations in fine-grained perception and reasoning. We further analyze failure cases and discuss directions for future model development. We release ChildBench and the evaluation code at a public anonymous URL: https://anonymous.4open.science/r/ChildBench-AF78.

Easy for Children, Hard for AI: The Limits of Multimodal LLMs in Early Childhood Learning

Large-scale map construction is foundational for critical applications such as autonomous driving and navigation systems.
Traditional large-scale map construction approaches mainly rely on costly and inefficient special data collection vehicles and labor-intensive annotation processes.
While existing satellite-based methods have demonstrated promising potential in enhancing the efficiency and coverage of map construction, they exhibit two major limitations: 
(1) inherent drawbacks of satellite data (e.g., occlusions, outdatedness)
and (2) inefficient vectorization from perception-based methods, resulting in discontinuous and rough roads that require extensive post-processing.
This paper presents a novel generative framework, UniMapGen, for large-scale map construction, offering three key innovations: 
(1) representing lane lines as discrete sequence and establishing an iterative strategy to generate more complete and smooth map vectors than traditional perception-based methods.
(2) proposing a flexible architecture that supports multi-modal inputs, enabling dynamic selection among BEV, PV, and text prompt, to overcome the drawbacks of satellite data.
(3) developing a state update strategy for global continuity and consistency of the constructed large-scale map.
UniMapGen achieves state-of-the-art performance on the OpenSatMap dataset. 
Furthermore, UniMapGen can infer occluded roads and predict roads
missing from dataset annotations.
Our code will be released.

UniMapGen: A Generative Framework for Large-Scale Map Construction from Multi-modal Data

Vision-Language Models (VLMs) have been applied to autonomous driving to support decision-making in complex real-world scenarios. However, their training on static, web-sourced image-text pairs fundamentally limits the precise spatiotemporal reasoning required to understand and predict dynamic traffic scenes. We address this critical gap with STRIDE-QA, a large-scale visual question answering (VQA) dataset for physically grounded reasoning from an ego-centric perspective. Constructed from 100 hours of multi-sensor driving data in Tokyo, capturing diverse and challenging conditions, STRIDE-QA is the largest VQA dataset for spatiotemporal reasoning in urban driving, offering 16 million QA pairs over 285K frames. Grounded by dense, automatically generated annotations including 3D bounding boxes, segmentation masks, and multi-object tracks, the dataset uniquely supports both object-centric and ego-centric reasoning through three novel QA tasks that require spatial localization and temporal prediction. Our benchmarks demonstrate that existing VLMs struggle significantly, achieving near-zero scores on prediction consistency. 
In contrast, VLMs fine-tuned on STRIDE-QA exhibit dramatic performance gains, achieving 55\% success in spatial localization and 28\% consistency in future motion prediction, compared to near-zero scores from general-purpose VLMs.
Therefore, STRIDE-QA establishes a comprehensive foundation for developing more reliable VLMs for safety-critical autonomous systems. Dataset and code will be released upon acceptance.

STRIDE-QA: Visual Question Answering Dataset for Spatiotemporal Reasoning in Urban Driving Scenes

Language models (LMs) are often adapted through supervised fine-tuning (SFT) to specialize their capabilities for downstream tasks. However, in typical scenarios where the fine-tuning data is limited, e.g., compared to pre-training, SFT can lead LMs to overfit, causing them to rely on spurious patterns within the target task or to compromise other broadly useful capabilities as a side effect of narrow specialization. In this paper, we propose Learning-from-the-Undesirable (LfU), a simple yet effective regularization scheme for SFT to mitigate overfitting issues when fine-tuning LMs with limited data. Specifically, we aim to regularize the fine-tuning process to favor solutions that are resilient to “undesirable” model updates, e.g., gradient ascent steps that steer the model toward undesirable behaviors. To this end, we propose a novel form of consistency regularization that directly aligns internal representations of the model with those after an undesirable update. By leveraging representation-level data augmentation through undesirable updates, LfU effectively promotes generalization under limited data. Our experiments on diverse LM downstream tasks show that LfU serves as an effective prior that enhances adaptability while preserving pretrained knowledge. For example, our LM from LfU achieves a 16.8% average improvement on math tasks compared to vanilla SFT on the same dataset, where the latter even leads to degraded performance on those tasks. Furthermore, LfU exhibits improved robustness to prompt variations, e.g., yielding a 92.1% lower standard deviation in output performances compared to SFT, highlighting its versatile effects.

Learning from the Undesirable: Robust Adaptation of Language Models Without Forgetting

The use of learned dynamics models, also known as world models, can improve the sample efficiency of reinforcement learning. Recent work suggests that the underlying causal graphs of such dynamics models are sparsely connected, with each of the future state variables depending only on a small subset of the current state variables, and that learning may therefore benefit from sparsity priors. Similarly, temporal sparsity, i.e. sparsely and abruptly changing local dynamics, has also been proposed as a useful inductive bias. In this work, we critically examine these assumptions by analyzing ground truth dynamics from a set of robotic reinforcement learning environments in the MuJoCo Playground benchmark suite, aiming to determine whether the proposed notions of state and temporal sparsity actually tend to hold in typical reinforcement learning tasks. We study (i) whether the causal graphs of environment dynamics are sparse, (ii) whether such sparsity is state-dependent, and (iii) whether local system dynamics change sparsely. Our results indicate that global sparsity is rare, but instead the tasks show local, state-dependent sparsity in their dynamics and this sparsity exhibits distinct structures, appearing in temporally localized clusters (e.g., during contact events) and affecting specific subsets of state dimensions. These findings challenge common sparsity prior assumptions in dynamics learning, emphasizing the need for grounded inductive biases that reflect the state-dependent sparsity structure of real-world dynamics.

Dynamic Sparsity: Challenging Common Sparsity Assumptions for Learning World Models in Robotic Reinforcement Learning Benchmarks

The more extensive access to time-series data, especially
for biomedical purposes, raises new methodological
challenges, particularly regarding missing values.
Functional linear discriminant analysis (FLDA) extends
Linear Discriminant Analysis (LDA)-mediated multiclass
classification and dimension reduction to data in the form
of fragmented observations of a univariate function. For
large multivariate and partially-observed data, there are
two challenges: (i) statistical dependencies between
different components of a multivariate function and (ii)
heterogeneous sampling times with missing features. We here
develop a multivariate version of FLDA, called MUDRA, to
tackle these challenges and describe a computationally
efficient expectation/conditional-maximisation (ECM)
algorithm to infer its parameters without any tensor
inversions. We assess its predictive power on the
“Articulary Words” dataset and show its improvement over
the state-of-the-art, especially in the case of missing
data. This advancement in dimension reduction of
multivariate functional data holds promise for enhancing
classification accuracy in scenarios like partially
observed short multivariate time series analysis.

Multivariate functional linear discriminant analysis for
partially-observed time series

Safety alignment instills in Large Language Models (LLMs) a critical capacity to refuse malicious requests. Prior works have modeled this refusal mechanism as a single linear direction in the activation space. We posit that this is an oversimplification that conflates two functionally distinct neural processes: the detection of harm and the execution of a refusal. In this work, we deconstruct this single representation into a Harm Detection Direction and a Refusal Execution Direction. Leveraging this fine-grained model, we introduce Differentiated Bi-Directional Intervention (DBDI), a new white-box framework that precisely neutralizes the safety alignment at critical layer. DBDI applies adaptive projection nullification to the refusal execution direction while suppressing the harm detection direction via direct steering. Extensive experiments demonstrate that DBDI outperforms prominent jailbreaking methods, achieving up to a 97.88% attack success rate on models such as Llama-2. By providing a more granular and mechanistic framework, our work offers a new direction for the in-depth understanding of LLM safety alignment.

Downloads

Next from AAAI 2026

Inductive Generative Recommendation via Retrieval-based Speculation

Similar lecture

Reward-Robust Reinforcement Learning in LLMs via Uncertainty Set Modeling

.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

Inductive Generative Recommendation via Retrieval-based Speculation

Similar lecture

Reward-Robust Reinforcement Learning in LLMs via Uncertainty Set Modeling

Downloads