论文检索

Yuhao Zhang, Ningkang Peng, Yafei Liu, Lin Li, Masaru Kitsuregawa, Yanhui Gu

The two-dimensional (2D) graph structure of a molecule encodes abundant latent property information. A well-designed molecular graph encoder can capture informative low-dimensional dense representations of molecules, which can subsequently be applied to a widerange of downstream tasks. To achieve fine-grained anddiscriminative molecular representations that capture localized structural information, we propose an novel atom-level adaptive receptive field encoder, enabling each atomic node in the molecular graph to dynamically adjust its receptive field size. To the best of our knowledge, we are the first to introduce an effective rank-guided pruning strategy for 2D molecular graphs.

Tianyu Zhang, I-Chao Shen, Haoran Xie

Anime hair design is crucial but challenging, as it conveys personality and emotion through stylized geometry and layered structure. In this work, we propose a sketch-guided approach for intuitive control of multimodal diffusion transformers (MMDiT) to generate semantically consistent anime hairstyles. We adopt a wisp-level flowline input integrated with a fine-tuned MMDiT to transfer hairstyles while preserving character identity. We believe that this fine-grained sketch control within the MMDiT framework may offer a promising path for structured anime hair editing.

Ryan Zhang, Herbert Woisetschläger

Real-world AI systems are tackling increasingly complex problems, often through interactions among Large Language Model (LLM) agents. When these agents develop inconsistent conventions, coordination can break down. Applications such as collaborative coding and distributed planning therefore require reliable, consistent communication, and scalability is a central concern as systems grow. We introduce Schema-Induced Games for Naming (SIGN), a naming game that examines how lightweight structure can steer convention formation. We compare schema-induced communication to unconstrained natural language and find faster convergence with up to 5.8× higher agreement. These results suggest that minimal structure can act as a simple control knob for efficient multi-agent coordination, pointing toward broader applications beyond the naming game.

Rongyu Zhang, Dingyuan Zhang, Haopeng Li

Proactive dialogue systems, which are designed to guide conversations toward predetermined goals. However, contemporary LLMs predominantly function as passive assistants, mechanically executing human instructions. A key challenge contributing to this limitation is the inherent difficulty in acquiring and annotating high-quality training data for proactive dialogue. Consequently, the scarcity of such data results in a notable deficiency in the proactive conversational capabilities of current LLMs.In this paper, we introduce PANDA (Proactive Agent-based Negotiation Dialogue Augmentation), a method designed to generate accurate, complex, and diverse proactive dialogue data for a challenging task—financial dispute mediation—where a LLM acts as the mediator. PANDA leverages a novel self-evolving synthesis process to manage a pool of user profiles and generate dialogues through structured interactions between multiple LLM-driven agents. To ensure data fidelity, we propose a comprehensive evaluation framework and build a two-level validation system combining automated and expert human verification. Our experiments demonstrate that an 8B-parameter model, trained on our synthesized dataset, achieves state-of-the-art results in the task's evaluation framework. Its performance rivals top closed-source models guided by heavily engineered prompts, even when provided with only essential information.

Hanzhang Yuan, Sheng Li

Estimating causal effects under network interference is challenging especially when edges are heterogeneous and nodes share latent dependencies. We study this realistic setting and propose MVDR, a targeted maximum likelihood (TMLE) framework that learns multi-view representations of covariates and exposure on heterogeneous networks while achieving double robustness: consistency holds if either the outcome model or the exposure density is correctly specified. MVDR supports multiple network interventions using only the observed network structure. On three semi-synthetic datasets, MVDR reduces intervention-level prediction error against baselines, and remains stable under misspecification.

Zhenyu Yu, Mohd Yamani Idna Idris, Pei Wang, Rizwan Qureshi

Quantitative remote sensing estimation is critical for environmental monitoring, providing continuous measures of vegetation indices, canopy height, and carbon stock. Traditional radiative-transfer models and empirical regressions require expert knowledge and generalize poorly, while deep learning methods remain task-specific. We propose SatelliteCalculator+, a DINOv3-powered multi-task foundation model for continuous regression of spectral and structural variables. The framework combines prompt-driven cross-attentive adapters with lightweight MLP decoders, enabling efficient dense prediction from frozen features. To overcome limited supervision, we synthesize over one million paired samples from SPOT 6/7 imagery using physically defined formulas. On the Open-Canopy dataset, SatelliteCalculator+ achieves competitive accuracy across eight ecological variables while reducing inference cost, demonstrating the promise of self-supervised transformers and scalable multi-task learning for large-scale Earth observation.

Ruteng Yu, Lukun Wang, Jiaming Pei

Remote sensing change detection (RSCD) is crucial for ur- ban monitoring, environmental protection, and disaster as- sessment, but small-sample scenarios often lead to overfitting and inaccurate predictions on unseen data. To address this, we propose GSAG-CDGAN, an end-to-end framework integrat- ing Selective Noise Augmentation (SNA) to mitigate overfit- ting, an Attention-Guided Adversarial Network (AGAN) to enhance structural consistency, and a Perceptual Loss Mod- ule (PLM) to preserve semantic consistency. Experiments on CDData-50 show that GSAG-CDGAN improves F1-Score from 0.6954 to 0.8851, with notable gains in Recall and IoU, demonstrating enhanced robustness under small-sample con- ditions. Further evaluation on the WHU-CD dataset yields an F1-Score of 0.9502, confirming strong cross-dataset general- ization and the method’s effectiveness in diverse scenarios.

Emily Yu

The identification of unique traits and behavior is essential to providing personalized intervention in individuals with Autism Spectrum Disorder. However, the limited personalized quantitative data with experts' annotations in autism research pose a fundamental challenge to train AI models for unique behavioral patten discovery. Multiple Instance Learning (MIL) has demonstrated promising results in medical domains, where annotations are only needed at the group level (i.e., a whole sequence) instead of individual data instances. It provides a cost-effective way to train statistical models with limited labeled data. Additionally, the rise of pretrained models have shown great success in improving the performance in few-shot learning scenarios. In this proof-of-concept study, we propose a novel framework that integrates a transformer encoder pre-trained on large-scale spatiotemporal data with MIL, for unique behavioral pattern detection from autistic individuals. Our results demonstrated the discrimination of individual-level autistic behavioral differences and the accurate classification of behaviors across distinct groups: typically developing (TD) and autistic (ASD). Beyond aggregate performance metrics, we highlight visual insights from temporal instance scores, revealing interpretable differences between individuals in their respective groups. These results show promising progress towards tools that can be used for personalized intervention for autistic individuals, and more interpretable AI diagnostics.

Byunghyuk Youn, Ohyun Jo

In this work, C2R-KD is proposed, applying a Complex-to-Real projection to map complex domain features into the real domain. C2R-KD mitigates complex-real domain mismatch to strengthen the representational capacity of the student model and further improves the knowledge distillation model performance through the hybrid distillation of features and logits simultaneously. Experimental result demonstrates higher accuracy than the conventional KD across all test environments.

Brandon Yee, Wilson Collins, Caden Wang, Mihir Tekal

Phase transitions in condensed matter systems traditionally require prior knowledge of order parameters for identification. We present Prometheus, a variational autoencoder framework for unsupervised discovery of phase transitions and order parameters in the two-dimensional Ising model without prior physical knowledge. Our approach combines convolutional neural networks with beta-variational autoencoders to learn compressed representations that naturally separate ordered and disordered phases. Experimental validation demonstrates automatic discovery of the order parameter with 0.85 correlation to theoretical magnetization and critical temperature detection within 0.27% of the theoretical value, achieving 89% improvement over principal component analysis while requiring no supervision.

Cai Yang, Kokil Jaidka, Subhayan Mukerjee, Yphtach Lelkes

We study when users end a session on X using high-resolution interaction logs from 215 US participants collected over four weeks. Sessions are defined via data-driven inter-activity gaps, and each session is encoded by fine-grained activity counts and duration (versus a simple activity ratio baseline). Fine-grained activity features substantially outperform the activity ratio baseline (C-index ≈ 0.76 vs. 0.62 for future sessions; 0.72 vs. 0.60 for unseen users), indicating that the composition of activity types is a strong predictor of disengagement. At the app level, we analyze retention over early adoption windows and find that the ratio of active activity in the first three days is most predictive of later usage. These results highlight session composition and early on-platform behavior as practical levers for forecasting and mitigating premature drop-off.

Yang Yan, Matthew W. Chen, Jiayi Lyu, Chen Zhao, Hao Gao, Zhong Chen

Radiotherapy (RT) is a cornerstone of cancer treatment. Following RT, patient-reported outcomes (PROs) collected via standardized questionnaires are crucial for monitoring patients' quality of life and side effects. However, traditional statistical and machine learning methods, which rely on structured numerical data, often fail to capture semantic meaning within patients' health status. To address this, we developed a novel framework using zero- and few-shot large language models (LLMs) to identify patients experiencing mild to severe depression. Furthermore, classification performance is enhanced through parameter-efficient fine-tuning. Experiments on a prostate cancer PRO dataset for depression have demonstrated that our fine-tuned LLMs consistently outperformed other baseline methods across key evaluation metrics.

Xiang Xue, Yatu Ji, Qing-Dao-Er-Ji Ren, Bao Shi, Min Lu, Nier Wu, Xufei Zhuang, Haiteng Xu, Gan-Qi-Qi-Ge Cha

Logit Knowledge Distillation has gained substantial research interest in recent years due to its simplicity and lack of requirement for intermediate feature alignment; however, it suffers from limited interpretability in its decision-making process. To address this, we propose implicit Clustering Distillation (iCD): a simple and effective method that mines and transfers interpretable structural knowledge from logits, without requiring ground-truth labels or feature-space alignment. iCD leverages Gram matrices over decoupled local logit representations to enable student models to learn latent semantic structural patterns. Extensive experiments on benchmark datasets demonstrate the effectiveness of iCD across diverse teacher-student architectures, with particularly strong performance in fine-grained classification tasks---achieving a peak improvement of +5.08% over the baseline.

Le Xu, Jiayu Chen

Offline reinforcement learning (RL) is vulnerable to real-world data corruption, with even robust algorithms failing under challenging observation and mixture corruptions. We posit this failure stems from data corruption creating sharp minima in the loss landscape, leading to poor generalization. To address this, we are the first to apply Sharpness-Aware Minimization (SAM) as a general-purpose, plug-and-play optimizer for offline RL. SAM seeks flatter minima, guiding models to more robust parameter regions. We integrate SAM into strong baselines for data corruption: IQL, a top-performing offline RL algorithm in this setting, and RIQL, an algorithm designed specifically for data-corruption robustness. We evaluate them on D4RL benchmarks with both random and adversarial corruption. Our SAM-enhanced methods consistently and significantly outperform the original baselines. Visualizations of the reward surface confirm that SAM finds smoother solutions, providing strong evidence for its effectiveness in improving the robustness of offline RL agents.

Pattaraphon Kenny Wongchamcharoen, Paul Glasserman

Large language models have shown great potential as forecasting tools in finance and economics, but backtesting performance is subject to look-ahead bias if the period overlaps with an LLM’s training window. Prompt-based attempts to avoid look-ahead bias require that LLMs understand chronology. We test LLMs’ ability to understand and enforce chronological order in three types of tasks: sorting randomly shuffled historical events; conditional sorting of events defined by some conditions; and anachronism detection based on intersections of multiple timelines. Our experiments use events that we first confirm are known to the LLM; this ensures that we test chronological understanding on an LLM’s pretrained internal knowledge. Across three LLM families— GPT-4.1 (standard), GPT-5 (hybrid-reasoning), and Claude 3.7 Sonnet (large-reasoning, with and without Extended Thinking), we find that performance degrades rapidly with problem complexity but improves greatly for reasoning models with test-time extended reasoning. These patterns are important for the real-time application of LLMs in finance.

Zuoyuehe Wang, Chicheng Ma, Pengpeng Chen, Lei Chai, Yongqiang Yang, Zhijun Chen, Jingzheng Li, Bing Li, Ting Wang

Adversarial training is an effective technique for enhancing the robustness of deep neural networks (DNNs). Prior research shows that misclassified examples influence final adversarial robustness much more than correctly classified examples. Ignoring this difference during training can hurt model performance. In crowdsourcing, varying annotator expertise causes noisy, inconsistent labels. As a result, it is hard to distinguish misclassified and correctly classified examples using only provided annotations. Thus, how to use the reliability and discrepancy between these example types to improve robustness within adversarial learning remains a critical but underexplored issue. In this work, we first explore how misclassified and correctly classified examples affect learning from crowds (LFC) in adversarial environments. Then, we formulate the problem of misclassification-aware robust learning from multiple human labelers as a bilevel min-max problem. After that, we introduce MALC, a new approach to make classifiers more robust to adversarial examples via iterative adversarial example generation and parameter estimation. We conduct an extensive evaluation of the proposed MALC, showing that MALC can outperform the state-of-the-art LFC methods in both white-box and black-box settings.

Yian Wang, Ye Qiao, Sitao Huang, Hyoukjun Kwon

We present APEX-Q, a flexible product quantization framework for compressing large language models. Unlike prior multi-codebook quantization methods with fixed partitions, APEX-Q supports arbitrary-dimensional tensor quantization, better capturing weight redundancy. It achieves performance on par with 4-bit and 8-bit baselines, enables post-training quantization without retraining, and reveals key trade-offs across subvector dimensions, codebook sizes, and hardware efficiency. APEX-Q thus provides a unified, hardware-friendly approach to scalable LLM deployment.

Li-Chiao Wang, Chung-Shou Liao, Wei Liu

Adversarial attacks remain a major challenge for deep learning models, as they can undermine both performance and reliability in practical applications such as image recognition. Although evolutionary algorithms (EAs) have proven effective in optimizing complex systems, their use for directly enhancing model robustness for adversarial defense has been limited. In this study, we introduce ResNet-GA, a method that applies evolutionary deep learning (EDL) to develop ResNet-like networks specifically designed to resist different forms of adversarial perturbations. The approach evolves network architectures with a genetic algorithm (GA), adapting the Residual Blocks at every stage in ResNet according to the needs of each dataset and attack type. Experimental results show that ResNet-GA strengthens model robustness beyond standard baselines, highlighting the value of iterative evolutionary design for building more dependable deep learning systems under various adversarial conditions.

Atul Teja Vellampalli, Yugan Kumar, Kai Zhe Tan, Kok Yang Tan, Vanessa Koh, Rahul Malhotra, Angelique W. M. Chan, David Matchar, Navrag B. Singh, Sai G. S. Pai

Falls are a major cause of injury and loss of independence among older adults, making prevention a critical priority for healthy aging. Early detection of fall risk through screening can enable timely interventions that reduce these adverse out-comes. Traditional clinical methods, such as using the history of falls and simple questionnaire-based screening, provide a quick and low-cost means of assessment but often have poor predictive accuracy and fail in presence of missing information. To support cost-effective screening and intervention, there is a need for tools that can accurately assess fall-risk in presence of missing information with better accuracy than cur-rent approaches. In this study, we developed a k-Nearest Neighbors (kNN) model that predicts whether an older adult will experience at least one fall within 12 months after baseline assessment, while simultaneously imputing missing data. Using data from 2,291 community-dwelling older adults in Singapore and 317 features spanning gait, cognition, physical activity, and comorbidities, our model achieved an AUC = 0.62 and F1 = 0.40, a significant improvement over the current clinical standard based solely on fall history (AUC ≈ 0.50). This model offers a more cost-effective screening tool for large-scale community deployment and highlights the feasibility of light-weight, imputation-aware models for practical fall-risk screening in aging populations.

Suchetan G. Uppur, Ashish Kumar, Akshay Agarwal

The Completely Automated Public Turing test to Tell Computers and Humans Apart (CAPTCHA) is widely deployed on the web as a security mechanism to distinguish humans from automated bots. However, their robustness is being challenged by the rapid advancements in AI, with models capable of near-human level character recognition rendering CAPTCHA obsolete. This research aims to systematically study the effect of multiple image corruptions, including elastic transformations, blur, noise, and occlusions, on human readability and automated solvers in text-based CAPTCHA recognition. We conduct experiments on multimodal large language models (MLLMs), a traditional deep learning-based optical character recognition (OCR) system, and human subjects. Using an existing CAPTCHA dataset and artificially corrupted versions, we analyze the recognition performance of AI models and humans, identifying vulnerabilities and patterns of robustness. The findings will contribute to a better understanding of CAPTCHA vulnerabilities and explore potential methods to increase the robustness of CAPTCHA in the era of advanced AI models.