AAAI 2026 Papers

Game UI development is essential to the game industry. However, the traditional workflow requires substantial manual effort to integrate pairwise UI and UX designs into a cohesive game user interface (GameUI). The inconsistency between the aesthetic UI design and the functional UX design typically results in mismatches and inefficiencies. To address the issue, we present an automatic system, AutoGameUI, for efficiently and accurately constructing GameUI. The system centers on a two-stage multimodal learning pipeline to obtain the optimal correspondences between UI and UX designs. The first stage learns the comprehensive representations of UI and UX designs from multimodal perspectives. The second stage incorporates grouped cross-attention modules with constrained integer programming to estimate the optimal correspondences through top-down hierarchical matching. The optimal correspondences enable the automatic GameUI construction. We create the GAMEUI dataset, comprising pairwise UI and UX designs from real-world games, to train and validate the proposed method. Besides, an interactive web tool is implemented to ensure high-fidelity effects and facilitate human-in-the-loop construction. Extensive experiments on the GAMEUI and RICO datasets demonstrate the effectiveness of our system in maintaining consistency between the constructed GameUI and the original designs. When deployed in the workflow of several mobile games, AutoGameUI achieves a 3× improvement in time efficiency, conveying significant practical value for game UI development.

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-DLite, a data distillation framework that enables more efficient pre-training by selectively removing noisy and redundant samples from large EEG datasets. EEG-DLite 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-DLite 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-DLite 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-DLite provides a scalable and practical path toward more effective and efficient physiological foundation modeling.

Recent progress in robotics and embodied AI is largely driven by Large Multimodal Models (LMMs). However, a key challenge remains underexplored: how can we advance LMMs to discover tasks that assist humans in open-future scenarios, where human intentions are highly concurrent and dynamic. In this work, we formalize the problem of Human-centric Open-future Task Discovery (HOTD), focusing particularly on identifying tasks that reduce human effort across plausible futures. To facilitate this study, we propose HOTD-Bench, which features over 2K real-world videos, a semi-automated annotation pipeline, and a simulation-based protocol tailored for open-set future evaluation. Additionally, we propose the Collaborative Multi-Agent Search Tree (CMAST) framework, which decomposes complex reasoning through a multi-agent system and structures the reasoning process through a scalable search tree module. In our experiments, CMAST achieves the best performance on the HOTD-Bench, significantly surpassing existing LMMs. It also integrates well with existing LMMs, consistently improving performance.

Generating behaviors that align with human expectations is a key requirement for human-robot collaboration. Potential behavior misalignment could lead to the robot performing actions with unanticipated, potentially dangerous side effects even while pursuing human goals. In this paper, we introduce a novel metric called Goal State Divergence (GSD) which quantifies the difference between the state a robot achieved in response to a human-specified goal and what the human expected. In cases where GSD cannot be directly calculated, we show how it can be approximated using maximal and minimal bounds. We then leverage GSD in our novel human-robot goal alignment design (HRGAD) problem, which identifies a minimal set of environment modifications that can reduce such mismatches. We show the effectiveness of our method in reducing the goal state divergence by empirically evaluating our approach on several planning benchmarks.

Self-interpretable models are increasingly valued for their inherent explainability. Among them, part-prototype networks stand out by mimicking human reasoning through the use of learned prototypes. However, their explanations often lack stability, becoming sensitive to subtle input perturbations. In this work, we propose Prototype in Imagery Network (PINet), a framework that improves the stability of prototype-based explanations. Rather than training on all possible input variations, which is computationally infeasible, PINet draws inspiration from visual mental imagery. Specifically, we incorporate empty inputs and apply coarse location guidance to simulate the human ability to imagine rough object features (a process akin to Phantasia). PINet mimics this process by incorporating empty inputs and applying coarse location guidance. These imagined, or uncertain, representations are contrasted with those derived from actual inputs (certain representations). We model the differences between the two by computing similarity at both the feature and prototype levels, allowing uncertainty to be explicitly encoded during prototype learning. Comprehensive evaluations on CUB-200-2011 and Stanford Cars demonstrate that PINet consistently achieves robust accuracy and localization, even under noisy conditions. These results represent the ability of PINet to produce stable and interpretable explanations under uncertainty.

Adaptive treatment strategies (ATS) are sequential decision-making processes that enable personalized care by dynamically adjusting treatment decisions in response to evolving patient symptoms. While reinforcement learning (RL) offers a promising approach for optimizing ATS, its conventional online trial-and-error learning mechanism is not permissible in clinical settings due to risks of harm to patients. Offline RL tackles this limitation by learning policies exclusively from historical treatment data, but its performance is often constrained by data scarcity—a pervasive challenge in clinical domains. To overcome this, we propose Treatment Stitching (TreatStitch), a novel data augmentation framework that generates clinically valid treatment trajectories by intelligently stitching segments from existing treatment data. Specifically, TreatStitch identifies similar intermediate patient states across different trajectories and stitches their respective segments. Even when intermediate states are too dissimilar to stitch directly, TreatStitch leverages the Schrödinger bridge method to generate smooth and shortest possible bridging trajectories that connect dissimilar states. By augmenting these synthetic trajectories into the original dataset, offline RL can learn from a more diverse dataset, thereby improving its ability to optimize ATS. Extensive experiments across multiple treatment datasets demonstrate the effectiveness of TreatStitch in enhancing offline RL performance. Furthermore, we provide a theoretical justification showing that TreatStitch maintains clinical validity by avoiding out-of-distribution transitions.

The growing demand for psychological support underscores the lack of high-quality counseling dialogue datasets, particularly in non-English contexts. We propose PGSim, a Path-Guided Simulation framework that mirrors real counseling processes—symptom description, problem identification, cause analysis, strategy planning, and iterative adjustment. PGSim models each user scenario as a fine-grained quadruple {Group, Psychological Problem, Problem Cause, Support Focus} and guides dialogue generation through expert-annotated strategy paths. Real counseling dialogues and expert-edited samples are used to fine-tune two language models: a Dialog Generator for strategy-aligned dialogue creation and a Dialog Modifier for expert-level refinement. After automated and human verification, we construct the Chinese Psychological support Dialogue Dataset (CPsDD), containing 68K dialogues across 13 groups, 16 problems, 13 causes, and 12 support focuses. We further present the Comprehensive Agent Dialogue Support System (CADSS), which integrates profiling, summarization, strategy planning, and empathetic response. Experiments on CPsDD and ESConv demonstrate that CADSS achieves state-of-the-art results on Strategy Prediction and Emotional Support Conversation tasks.

Livestreaming has become increasingly prevalent in modern visual communication, where automatic camera quality tuning is essential for delivering superior user Quality of Experience (QoE). Such tuning requires accurate blind image quality assessment (BIQA) to guide parameter optimization decisions. Unfortunately, the existing BIQA models typically only predict an overall coarse-grained quality score, which cannot provide fine-grained perceptual guidance for precise camera parameter tuning. To bridge this gap, we first establish FGLive-10K, a comprehensive fine-grained BIQA database containing 10,185 high-resolution images captured under varying camera parameter configurations across diverse livestreaming scenarios. The dataset features 50,925 multi-attribute quality annotations and 19,234 fine-grained pairwise preference annotations. Based on FGLive-10K, we further develop TuningIQA, a fine-grained BIQA metric for livestreaming camera tuning, which integrates human-aware feature extraction and graph-based camera parameter fusion. Extensive experiments and comparisons demonstrate that TuningIQA significantly outperforms state-of-the-art BIQA methods in both score regression and fine-grained quality ranking, achieving superior performance when deployed for livestreaming camera tuning.

Reinforcement Learning from Human Feedback (RLHF) is a methodology that aligns agent behavior with human preferences by integrating human feedback into the agent's training process. We introduce a possible framework that employs passive Brain-Computer Interfaces (BCI) to guide agent training from implicit neural signals. We present and release a novel dataset of functional near-infrared spectroscopy (fNIRS) recordings collected from 25 human participants across three domains: a Pick-and-Place Robot, Lunar Lander, and Flappy Bird. We train classifiers to predict levels of agent performance (optimal, sub-optimal, or worst-case) from windows of preprocessed fNIRS feature vectors, achieving an average F1 score of 67% for binary classification and 46% for multi-class models averaged across conditions and domains. We also train regressors to predict the degree of deviation between an agent's chosen action and a set of near-optimal policies, providing a continuous measure of performance. We evaluate cross-subject generalization and demonstrate that fine-tuning pre-trained models with a small sample of subject-specific data increases average F1 scores by 17% and 41% for binary and multi-class models, respectively. Our work demonstrates that mapping implicit fNIRS signals to agent performance is feasible and can be improved, laying the foundation for future brain-driven RLHF systems.

Inferring humans' private valuations for goods from their observed market behavior is essential for evaluating market efficiency and improving trading mechanism design. A core challenge lies in uncovering the human decision function that maps private valuations and observed market states to actions. In complex market settings where humans make sequential decisions in stochastic environments, neural networks offer the flexibility to model this decision function. However, training them without access to private valuations or environment dynamics remains challenging. We tackle this challenge and study how to infer heterogeneous human valuations from offline decision data in continuous double auctions. We propose learning the decision function via risk‑sensitive utility maximization. First, we train a generative model on offline bid and ask data to simulate individual trading behavior. Using this generative model, we instantiate simulated markets composed of randomly generated buyers and sellers. Second, we introduce an agent into these simulated markets and use reinforcement learning to learn a risk-sensitive utility-maximizing decision function for the agent. Third, we formulate a bilevel optimization to jointly recover private valuations and risk preference parameters. Our extensive experiments on a large‑scale continuous double auction dataset demonstrate that our framework significantly reduces errors in inferring real human valuations.

Explanatory Interactive Learning (XIL) is a powerful interactive learning framework designed to enable users to customize and correct AI models by interacting with their explanations. In a nutshell, XIL algorithms select a number of items on which an AI model made a decision (e.g., images and their tags) and present them to users, together with corresponding explanations (e.g., image regions that drive the model’s decision). Then, users supply corrective feedback for the explanations, which the algorithm uses to improve the model. Despite showing promise in debugging tasks, recent studies have raised concerns that explanatory interaction may trigger order effects, a well-known cognitive bias in which the sequence of presented items influences users’ trust and, critically, the quality of their feedback. We argue that these studies are not entirely conclusive, as the experimental designs and tasks employed differ substantially from common XIL use cases, complicating interpretation. To clarify the interplay between order effects and explanatory interaction, we ran a larger-scale user study (n = 713 total) designed to mimic common XIL tasks. Specifically, we assessed order effects both within and between debugging sessions by manipulating the order in which correct and wrong explanations are presented to participants. Order effects had a limited but significant impact on users’ agreement with the model (i.e., a behavioral measure of their trust), and only when examined within debugging sessions, not between them. The quality of users’ feedback reached satisfactory levels overall, with order effects exerting only a small and inconsistent influence both within and between sessions. Overall, our findings suggest that order effects do not pose a significant issue to the successful employment of XIL approaches. More broadly, our work contributes to the ongoing efforts for understanding human factors in AI.

Consensus decision-making uses crowd responses (usually from non-experts) to questions to reach a consensus answer based on human-machine collaboration. The crucial point is dynamic, which should not only enable rapid self-iteration toward the correct answer through crowd workers' responses but also adaptively suggest the next most valuable question(s) to accelerate the integration of the answer. However, existing methods reach consensus using either offline data or fixed question search structures, thereby largely sidestepping this dynamic nature. In response, we propose a bilevel optimization-based human-machine collaboration (BiO-HMC), which explores an inner & outer-level optimization to enable effective answer integration and efficient question selection. The resulting optimization problem is intractable because there is no closed-form expression in the inner-level optimization. We employ a gradient-based method and guarantee the method's theoretical convergence. Experimental results on synthetic and real-world datasets demonstrate the effectiveness and efficiency of the BiO-HMC model, i.e., achieving the highest confidence in the correct answer with the lowest labor cost.

Ensuring alignment with human values is essential for modern large language models (LLMs), especially amid growing concerns around AI safety and social impact. Yet achieving such alignment remains challenging due to the limited, noisy, and often conflicting nature of human feedback from diverse annotators. Most existing approaches, such as Direct Preference Optimization (DPO), assume consistent and conflict-free supervision, overlooking the ambiguity, inconsistency, and value trade-offs inherent in real-world preferences—often leading to reduced robustness and exclusion of minority views. To address this, we propose FGD-Align, a novel pluralistic alignment framework grounded in Fuzzy Group Decision-Making theory. Our approach rigorously models and aggregates human preferences while retaining the complexity of real-world value trade-offs. Unlike traditional methods that rely on coarse-grained preference pairs, FGD-Align introduces fuzzy preference modeling via triangular fuzzy numbers to capture nuanced, multi-criteria human judgments. We further develop a new training objective, Probabilistic Fuzzy DPO, which incorporates fuzzy preference strength as adaptive loss weights and gradient filters, enhancing robustness to ambiguity and inconsistency in feedback. Comprehensive experiments demonstrate that FGD-Align consistently outperforms both DPO variants and advanced preference aggregation methods in terms of preference accuracy and robustness to ambiguity. It achieves superior alignment stability and better preserves minority preferences, all with minimal computational overhead. Our work bridges the gap between algorithmic tractability and the nuanced landscape of human values, enabling more scalable, inclusive, and socially-aware AI alignment.

While large language model (LLM)-based user profiling offers significant potential for personalization, most existing approaches rely on empirical heuristics and lack grounding in the psychological mechanism that drive human behavior. In this paper, we introduce TRIPLE (Theory-guided Reasoning for Intent and habIt Profiling with LLMs for pErsonalization), a novel framework that systematically integrates dual-process theory from social psychology into LLM-based user modeling. TRIPLE (1) constructs a habitual behavior profile by identifying repeated patterns over time to model automatic responses; (2) builds an intentional behavior profile by inferring user attitudes, subjective norms and perceived behavioral control based on the Theory of Planned Behavior (TPB); and (3) generates behavioral rationale that reveal the interaction between habitual and intentional processes to predict user behavior in context-specific situations. We evaluate TRIPLE on five personalization tasks from the LaMP benchmark using multiple open-source LLMs. Results show that TRIPLE consistently outperforms existing in-context learning methods, with especially pronounced gains on complex generative tasks such as headline and title generation. Qualitative analyses further demonstrate that the profiles and reasoning paths generated by TRIPLE provide interpretable and psychologically grounded explanations of user behavior. These findings provide strong evidence that incorporating validated behavioral theories into LLM-based personalization enhances both predictive performance and interpretability, paving a way for theory-driven, socio-cognitively informed user modeling.

Graph Transformer shows remarkable potential in brain network analysis due to its ability to model graph structures and complex node relationships. Most existing methods typically model the brain as a flat network, ignoring its modular structure, and their attention mechanisms treat all brain region connections equally, ignoring distance-related node connection patterns. However, brain information processing is a hierarchical process that involves local and long-range interactions between brain regions, interactions between regions and sub-functional modules, and interactions among functional modules themselves. This hierarchical interaction mechanism enables the brain to efficiently integrate local computations and global information flow, supporting the execution of complex cognitive functions. To address this issue, we propose BrainHGT, a hierarchical Graph Transformer that simulates the brain’s natural information processing from local regions to global communities. Specifically, we design a novel long-short range attention encoder that utilizes parallel pathways to handle dense local interactions and sparse long-range connections, thereby effectively alleviating the over-globalizing issue. To further capture the brain’s modular architecture, we designe a prior-guided clustering module that utilizes a cross-attention mechanism to group brain regions into functional communities and leverage neuroanatomical prior to guide the clustering process, thereby improving the biological plausibility and interpretability. Experimental results indicate that our proposed method significantly improves performance of disease identification, and can reliably capture the sub-functional modules of the brain, demonstrating its interpretability.

The recent DeepSeek-R1 has showcased the emergence of reasoning capabilities in large language models (LLMs) through reinforcement learning (RL) with rule-based rewards. Despite its success in language tasks, its application in multimodal domains, particularly in graphic user interface (GUI) agent tasks, remains under-explored. To address this gap, we propose UI-R1, the first framework to investigate how rule-based RL can enhance the reasoning capabilities of multimodal large language models (MLLMs) for GUI action prediction tasks. UI-R1 introduces a novel rule-based action reward scheme, enabling model optimization via policy-based algorithms such as Group Relative Policy Optimization (GRPO). To further improve efficiency at inference time, we present UI-R1-Efficient, a two-stage training paradigm that reduces reasoning length while boosting overall performance. In addition, we construct a compact yet high-quality dataset containing 2K challenging tasks across five prevalent mobile device action types. Experiments show that our proposed models (e.g., UI-R1-3B) achieve substantial improvements over the base model (Qwen2.5-VL-3B) on both in-domain (ID) and out-of-domain (OOD) tasks, with average accuracy gains of 18.3% on ScreenSpot, 6.0% on ScreenSpot-Pro, and 10.9% on ANDROIDCONTROL. Moreover, our efficient versions deliver competitive performance compared to considerably larger state-of-the-art models, underscoring the potential of reinforcement learning to advance GUI control and paving the way for future research in Human-Computer Interaction (HCI).

Space computing devices expand handwritten input from two-dimensional screens into three-dimensional space, providing an unrestricted interactive experience. Due to the high degree of freedom and lack of tactile feedback in in-air handwriting, handwritten characters not only become less legible but also lose the writer's personal style. This paper proposes a method for reconstructing discrete in-air handwriting using continuous diffusion models, capturing the writing process and style from a small number of user-provided handwritten tracks and images, to restore the legibility of characters and mimics the writer's style. We represent handwritten track data in binary form and model it with continuous diffusion models, recovering discrete handwritten track data through threshold processing. Our approach reconstructs in-air handwritten characters in two stages. During the content preservation phase, we propose a partial noise injection strategy based on reference sequence modeling, using the content information of the original character as a guiding condition to maintain content consistency in handwritten character. In the style aggregation phase, we adaptively fuse the visual style of the handwritten in the image modality with the dynamic writing process in the sequence modality, overcoming issues of insufficient style capture due to noise interference in the backward process. Qualitative and quantitative experiments demonstrate the superiority of our method.

Brain decoding aims to reconstruct video from brain signals. Existing brain decoding frameworks are primarily built on a subject-dependent paradigm, which requires large amounts of brain data for each subject. However, the expensive cost of collecting brain-video data causes severe data scarcity for brain decoding. Although some cross-subject methods being introduced, they often exhibit an excessive preoccupation with subject-invariant information while neglecting subject-specific information, resulting in slow fine-tune-based adaptation strategy. To achieve fast and data-efficient new subject adaptation, we propose **MindCross**, a novel cross-subject brain decoding framework. MindCross's *N* specific encoders and one shared encoder are designed to extract subject-specific and subject-invariant information, respectively. Additionally, a Top-*K* collaboration module is adopted to enhance new subject decoding with the knowledge learned from previous subjects' encoders. Extensive experiments on fMRI/EEG-to-video benchmarks demonstrate MindCross's efficacy and efficiency of cross-subject decoding and new subject adaptation using only one model. Code of our framework will be released upon publication.

In recent years, electroencephalography (EEG)-based visual decoding research has become a key direction for revealing brain processing mechanisms and realizing brain-computer interfaces. This emerging field has attracted extensive attention in the fields of brain science, cognitive neuroscience, and artificial intelligence. Among various approaches, contrastive learning has demonstrated strong performance in aligning multi-modal data, effectively enabling unified representations across modalities. However, during human visual perception, images are often subject to varying degrees of blurring due to the uneven distribution of retinal photoreceptor cells and the limited speed of lens accommodation. To address the mismatch between EEG and visual representations, we propose a novel visual decoding framework inspired by human perceptual blurring. Specifically, multi-level Gaussian blurring is applied to the visual stimuli to simulate human visual characteristics, followed by a feature selection module to construct robust visual representations. For EEG decoding, we design a lightweight and efficient network employing positively constrained spatial convolutions to identify channels associated with visual processing. The EEG and visual features are then aligned using contrastive learning. We evaluate the proposed framework on the Things-EEG dataset. Experimental results show significant improvements in the zero-shot brain-to-image retrieval task, achieving a top-1 accuracy of 80% and a top-5 accuracy of 96.9%, surpassing previous state-of-the-art methods by margins of 29.1% and 17.2%, respectively. These findings highlight the potential of incorporating perceptual properties into EEG-based visual decoding.