Singapore

Modern vision-language models (VLMs) deliver impressive predictive accuracy yet offer little insight into &#39;why&#39; a decision is reached, frequently hallucinating facts, particularly when encountering out-of-distribution data. Neurosymbolic frameworks address this by pairing black-box perception with interpretable symbolic reasoning, but current methods extract their symbols solely from task labels, leaving them weakly grounded in the underlying visual data. In this paper, we introduce a multi-agent system - Concept-RuleNet that reinstates visual grounding while retaining transparent reasoning. Specifically, a multimodal concept generator first mines discriminative visual concepts directly from a representative subset of training images. Next, these visual concepts are utilized to condition symbol discovery, anchoring the generations in real image statistics and mitigating label bias. Subsequently, symbols are composed into executable first-order rules by a large language model reasoner agent - yielding interpretable neurosymbolic rules. Finally, during inference, a vision verifier agent quantifies the degree of presence of each symbol and triggers rule execution in tandem with outputs of black-box neural models, predictions with explicit reasoning pathways. Experiments on five benchmarks, including two challenging medical-imaging tasks and three underrepresented natural-image datasets, show that our system augments state-of-the-art neurosymbolic baselines by an average of 5% while also reducing the occurrence of hallucinated symbols in rules by up to 50%.

AAAI 2026

Concept-RuleNet: Grounded Multi-Agent Neurosymbolic Reasoning in Vision Language Models

Modern vision-language models (VLMs) deliver impressive predictive accuracy yet offer little insight into 'why' a decision is reached, frequently hallucinating facts, particularly when encountering out-of-distribution data. Neurosymbolic frameworks address this by pairing black-box perception with interpretable symbolic reasoning, but current methods extract their symbols solely from task labels, leaving them weakly grounded in the underlying visual data. In this paper, we introduce a multi-agent system - Concept-RuleNet that reinstates visual grounding while retaining transparent reasoning. Specifically, a multimodal concept generator first mines discriminative visual concepts directly from a representative subset of training images. Next, these visual concepts are utilized to condition symbol discovery, anchoring the generations in real image statistics and mitigating label bias. Subsequently, symbols are composed into executable first-order rules by a large language model reasoner agent - yielding interpretable neurosymbolic rules. Finally, during inference, a vision verifier agent quantifies the degree of presence of each symbol and triggers rule execution in tandem with outputs of black-box neural models, predictions with explicit reasoning pathways. Experiments on five benchmarks, including two challenging medical-imaging tasks and three underrepresented natural-image datasets, show that our system augments state-of-the-art neurosymbolic baselines by an average of 5% while also reducing the occurrence of hallucinated symbols in rules by up to 50%.

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.<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.

Time series analysis has progressed from traditional
autoregressive models to deep learning, Transformers, and
foundation models (FMs), including large language models
(LLMs) and large vision models (LVMs). These advances have
expanded model design possibilities and, notably, enabled
time series problem-solving across multiple modalities,
greatly enhancing downstream applications in domains such
as climate, agriculture, and healthcare. In this talk, I
will provide an overview of recent developments in large
FMs for time series, highlighting frameworks for
transferring knowledge from other modalities to time
series, and identifying the advantages of LVMs in
cross-modal knowledge transfer over LLMs. I will then delve
into our recent research on LVMs for time series,
discussing (1) mainstream techniques for imaging time
series (i.e., transforming time series to images); (2)
empirical findings on LVMs' effectiveness and limitations
in time series modeling; and (3) integration of LVMs in
multimodal frameworks for time series encoding. Finally,
this talk will conclude with an application of LVMs in
brain time series learning in neuroscience. The aim of the
talk is to review state-of-the-art AI techniques for time
series, highlight unique challenges, and share our recent
findings in this promising area.

Cross-Modal Knowledge Transfer in Time Series AI via Large Vision Models

Artificial Intelligence Generated Content (AIGC) assisting image production triggers academic controversy widely in journalism, while attracts increasingly consideration by media agencies. Key issues involve misinformation, authenticity, semantic fidelity and interpretability. Now, Most AIGC tools operate as opaque “black boxes,” which make them difficult to meet the dual demands of content accuracy and semantic alignment in journalistic contexts, even cause ethical, sociotechnical and trusting dilemma. In this paper, the authors explore feasible pathways for controllable image production in the typical scenario of journalism – special coverage, and conduct two experiments based on coverage projects of China’s top media agency: (1) Experiment 1 focuses on cross-platform adaptability by applying standardized prompts to three coverage scenes, showing the significant disparities in semantic alignment, cultural specificity, and visual realism, which are largely influenced by training corpus bias and platform-level filtering policies. (2) In experiment 2, a human-in-the-loop modular pipeline is developed that combines high-precision segmentation (SAM, GroundingDINO), semantic alignment (BrushNet) and style regulating (Style-LoRA, Prompt-to-Prompt). It ensures the editorial fidelity by CLIP-based semantic scoring, NSFW/OCR/YOLO-based content filtering, and the embedding of verifiable content credentials. The traceable deployment of modular pipeline keeps the consistency of semantic representation and interpretation when producing images. Consequently, the authors propose a human-AI collaboration mechanism of AIGC assisted image production for special coverage, and corresponding evaluation principles is suggested in three main criteria: Character Identity Stability (CIS), Cultural Expression Accuracy (CEA) and User-Public Appropriateness (U-PA). This paper calls for systemic upgrades involving human-in-the-loop modelling, stylistic co-creation, and trust labelling mechanisms, and provides a practical framework for controllable and reliable AIGC deployment in image production of journalism.

A Human-AI Collaboration Mechanism Study on AIGC Assisted Image Production for Special Coverage

Diffusion models have demonstrated remarkable success in image generation, yet a persistent challenge remains: the bias between model predictions and the target distribution. In this paper, we propose a Bidirectional Noise Injection framework for enhancing diffusion models, implemented via Coordinated Input-Output Perturbation (CIOP). Our approach mitigates this bias by randomly applying synchronized noise injection to both the model inputs and the prediction targets during the training stage. This stochastic, synchronized noise injected acts as a smoothing mechanism that effectively reduces the 2-Wasserstein distance between the predicted and target distributions, as substantiated by our theoretical analysis based on optimal transport theory. Extensive experiments on multiple benchmark datasets and various generative tasks demonstrate that our method improves generation quality and training efficiency without incurring additional computational cost. Furthermore, the design of CIOP enables seamless integration with existing diffusion model improvements and advanced frameworks, thereby broadening its applicability. These results highlight the potential of Bidirectional Noise Injection via CIOP to alleviate bias in diffusion-based generative models across a wide range of settings.

Bidirectional Noise Injection: Enhancing Diffusion Models via Coordinated Input-Output Perturbation

Progress in multimedia technology has mitigated real-world data incompleteness and provided a versatile platform for multi-view learning.
Among existing research, graph-based multi-view learning has achieved notable success.
However, prior studies always immerse in comprehensive collaboration across all views and nodes to pursue consistency and complementary, which ignore the negative contribution of nodes from low-quality views.
To overcome the above limitation, we explore node behavior selection in multi-view dynamic modeling and propose the knowledge-aware multi-view state space model.
Specifically, nodes autonomously select either activation sequences or static sequences according to their current knowledge.
In the former, we design the mask-based attention mechanism to capture the dynamics of node behaviors. 
In the latter, we construct a history pool and simulate synaptic signals to regulate the behavioral distribution of nodes. 
Moreover, the proposed model provides a directional inter-view diffusion equation that selectively propagates information to alleviate interference from low-quality nodes across views.
Extensive experiments demonstrate that the proposed model outperforms baselines on multiple benchmarks and achieves significant performance improvement.

From Static to Active: Knowledge-Aware Node State Selection in Multi-view Graph Learning

Instruction tuning plays a critical role in enhancing the performance and efficiency of Large Language Models (LLMs). Its success depends not only on the quality of the instruction data but also on the inherent capabilities of the LLM itself. Some studies suggest that even a small amount of high-quality data can achieve instruction fine-tuning results that are on par with, or even exceed, those from using a full-scale dataset. However, rather than focusing solely on calculating data quality scores to evaluate instruction data, there is a growing need to select high-quality data that maximally enhances the performance of instruction tuning for a given LLM. In this paper, we propose the Model Instruction Weakness Value (MIWV) as a novel metric to quantify the importance of instruction data in enhancing model's capabilities. The MIWV metric is derived from the discrepancies in the model’s responses when using In-Context Learning (ICL), helping identify the most beneficial data for enhancing instruction tuning performance. Our experimental results demonstrate that selecting only the top 1% of data based on MIWV can outperform training on the full dataset. Furthermore, this approach extends beyond existing research that focuses on data quality scoring for data selection, offering strong empirical evidence supporting the effectiveness of our proposed method.

Importance-Aware Data Selection for Efficient LLM Instruction Tuning

Recent advances in generative artificial intelligence have enabled the creation of highly realistic image forgeries, raising significant concerns about digital media authenticity. While existing detection methods demonstrate promising results on benchmark datasets, they face critical limitations in real-world applications. First, existing detectors typically fail to detect semantic inconsistencies with the person’s identity, such as implausible behaviors or incompatible environmental contexts in given images. Second, these methods rely heavily on low-level visual cues, making them effective for known forgeries but less reliable against new or unseen manipulation techniques. To address these challenges, we present a novel personalized vision-language model (VLM) that integrates low-level visual artifact analysis and high-level semantic inconsistency detection. Unlike previous VLM-based methods, our approach avoids resource-intensive supervised fine-tuning that often struggles to preserve distinct identity characteristics. Instead, we employ a lightweight method that dynamically encodes identity-specific information into specialized identifier tokens. This design enables the model to learn distinct identity characteristics while maintaining robust generalization capabilities. We further enhance detection capabilities through a lightweight detection adapter that extracts fine-grained information from shallow features of the vision encoder, preserving critical low-level evidence. Comprehensive experiments demonstrate that our approach achieves 94.25% accuracy and 94.08% F1 score, outperforming both traditional forgery detectors and general VLMs while requiring only 10 extra tokens.

Identity-Aware Vision-Language Model for Explainable Face Forgery Detection

Lithium-ion (Li-ion) batteries are the major type of battery used in a variety of everyday applications, including electric vehicles (EVs), mobile devices, and energy storage systems. Predicting the Remaining Useful Life (RUL) of Li-ion batteries is crucial for ensuring their reliability, safety, and cost-effectiveness in battery-powered systems. The materials used for the battery cathodes and their designs play a significant role in determining the degradation rates and RUL, as they lead to distinct electrochemical reactions. Unfortunately, RUL prediction models often overlook the cathode materials and designs to simplify the model-building process, ignoring the effects of these electrochemical reactions. Other reasons are that specifications related to battery materials may not always be readily available, and a battery might consist of a mix of different materials. As a result, the predictive models that are developed often lack generalizability. To tackle these challenges, this paper proposes a novel material-based Mixture-of-Experts (MoE) approach for predicting the RUL of batteries, specifically addressing the complexities associated with heterogeneous battery chemistries. The MoE is integrated into a probabilistic framework, called \textit{Multiple Non-crossing Quantile Mixture-of-Experts for Probabilistic Prediction (RUL-QMoE)}, which accommodates battery operational conditions and enables uncertainty quantification. The RUL-QMoE model integrates specialized expert networks for five battery types: LFP, NCA, NMC, LCO, and NMC-LCO, within a gating mechanism that dynamically assigns relevance based on the battery's input features. Furthermore, by leveraging non-crossing quantile regression, the proposed RUL-QMoE produces coherent and interpretable predictive distributions of the battery's RUL, enabling robust uncertainty quantification in the battery's RUL prediction. Trained on seven real-world datasets, the proposed RUL-QMoE achieves strong predictive performance across all battery types, with MAE = 65 (cycles), MAPE = 9.59\%, RMSE = 100 (cycles), and $R^2=96.84\%$. Compared to traditional models like XGBoost, Random Forest, CNN, and LSTM, the proposed RUL-QMoE model consistently delivers lower RMSE and superior probabilistic insights, including survival probabilities and prediction intervals. The model has been integrated into our Battery AI platform in collaboration with Toyota Motor Engineering \& Manufacturing North America, Inc., as part of a broader Battery Foundation Model initiative. This RUL-QMoE model will serve future Toyota EVs' users and battery system designers.

RUL-QMoE: Multiple Non-crossing Quantile Mixture-of-Experts for Probabilistic Remaining Useful Life Predictions of Varying Battery Materials

Diffusion-based tabular data synthesis models have yielded promising results. However, we observe that when the data dimensionality increases, existing models tend to degenerate and may perform even worse than simpler, non-diffusion-based models. This is because limited training samples in high-dimensional space often hinder generative models from capturing the distribution accurately. To mitigate the insufficient learning signals and to stabilize training under such conditions, we propose CtrTab, a condition-controlled diffusion model that injects perturbed ground-truth samples as auxiliary inputs during training. This design introduces an implicit $L_2$ regularization on the model’s sensitivity to the control signal, improving robustness and stability in high-dimensional, low-data scenarios. Experimental results across multiple datasets show that CtrTab outperforms state-of-the-art models, with a performance gap in accuracy over 90% on average.

Towards Synthesizing High-Dimensional Tabular Data with Limited Samples

Designing effective algorithmic components remains a fundamental obstacle in tackling NP-hard combinatorial optimization problems (COPs), where solvers often rely on carefully hand-crafted strategies. Despite recent advances in using large language models (LLMs) to synthesize high-quality components, most approaches restrict the search to a single element—commonly a heuristic scoring function—thus missing broader opportunities for innovation. We introduce a broader formulation of solver design as a multi-strategy optimization problem, which seeks to jointly improve a set of interdependent components under a unified objective. To address this, we propose MOTIF—Multi-strategy Optimization via Turn-based Interactive Framework—a novel framework based on Monte Carlo Tree Search that facilitates turn-based optimization between two LLM agents. At each turn, an agent improves one component by leveraging the history of both its own and its opponent’s prior updates, promoting both competitive pressure and emergent cooperation. This structured interaction broadens the search landscape and encourages the discovery of diverse, high-performing solutions. Experiments across multiple COP domains show that MOTIF consistently outperforms state-of-the-art methods, highlighting the promise of turn-based, multi-agent prompting for fully automated solver design.

MOTIF: Multi-strategy Optimization via Turn-based Interactive Framework

We study hybrid execution in multi-agent reinforcement learning (MARL), a paradigm where agents aim to complete cooperative tasks with arbitrary communication levels at execution time by taking advantage of information-sharing among the agents. Under hybrid execution, the communication level can range from a setting in which no communication is allowed between agents (fully decentralized), to a setting featuring full communication (fully centralized), but the agents do not know beforehand which communication level they will encounter at execution time. We contribute MARO, an approach that makes use of an auto-regressive predictive model, trained in a centralized manner, to estimate missing agents' observations at execution time. We evaluate MARO on standard scenarios and extensions of previous benchmarks tailored to emphasize the impact of partial observability in MARL. Experimental results show that our method consistently outperforms relevant baselines, allowing agents to act with faulty communication while successfully exploiting shared information.

Premium content

Next from AAAI 2026

Cross-Modal Knowledge Transfer in Time Series AI via Large Vision Models

Stay up to date with the latest Underline news!

PRESENTATIONS

CONFERENCES

COMPANY

RESOURCES