论文检索

Malak Mahdy, Carlos Rubio-Medrano

Autonomous computer-use agents (ACUAs) enable end-to-end computer operation with human-like capabilities, executing commands across applications and making independent decisions. However, their real-world effectiveness and security remain largely untested. A systematic evaluation of ACUAs from Anthropic, OpenAI, and open-source projects categorized them into full computer access and browser-based agents. Findings reveal substantial limitations, with success rates dropping as low as 28% in some cases. Additionally, a 100% rate of unauthorized software installation was observed in certain tasks. The agents also demonstrated susceptibility to prompt injection attacks. The impact of varied prompting strategies on performance was also examined. In response to these weaknesses, a new agent framework designed to address these limitations is proposed. This work bridges agentic AI, human-computer interaction (HCI), and security to address the observed limitations of ACUAs, prioritizing both capability and safety.

Al Hasib Mahamud, Yllias Chali

Question generation is the task of natural language processing where the goal is to generate fluent, grammatically correct, error-free questions based on a given input context and optionally an answer. Multi-hop question generation is a more complex task compared to traditional single-hop question generation, as it requires reasoning over multiple information from multiple input contexts in generating multi-hop questions. In this paper, we have addressed the challenge of building a multi-hop question generation system by combining the knowledge graphs with large language models. We have designed a framework KG4QG (Knowledge Graph for Question Generation), where knowledge graphs are generated from the input contexts. For the knowledge graph embedding, we have used Graph Attention Network, and for input text embedding, we have leveraged Sentence Transformer. Finally, we apply BART and T5 models as Large Language Models to generate multi-hop questions from our proposed model. Using HotpotQA dataset to evaluate the performance of our KG4QG framework, our proposed methodology has shown an enhancement of performance over the previous methodologies.

Christopher MacDowell, Sarah Setiawan, Carol Jim, Ahmed Salem

Machine Learning (ML) models have significant potential across research and industry to enable data-driven insights and decision-making. Their performance relies on input data quality, but real-world datasets often contain imperfections, making data preprocessing essential yet time-consuming. Our research proposes a proof-of-concept model using Generative Artificial Intelligence (GenAI) to analyze and transform data for supervised ML classification. The results from the GenAI models will be compared with traditionally preprocessed data to evaluate effectiveness. Preliminary results indicate that incorporating GenAI models into the preprocessing pipeline show potential in improving ML's classification performance.

Tianzi Ma, Yulin Wu, Hang Ren, Xiaozhen Sun, Shuhan Qi, Xuan Wang

Accurately modeling opponent behaviors and integrating strategy are key challenges for multi-issue automated negotiation. Existing approaches often isolate preference learning or trend prediction and lack a unified cognitive structure with coordinated reasoning. This paper proposes a BDI (Belief-Desire-Intention)-based opponent modeling and strategy generation framework. The framework analyzes opponent responses (Belief), predicts preference weights and the utility function (Desire), and infers utilities of future offers (Intention). Building on these predictions, we design a responsive strategy, enabling gradual concessions and balanced outcomes. Our main contributions are: D-MBUE in the Desire module, I-DABI in the Intention module, and the BDI Negotiator on top of the modeling modules. Experiments on 45 standard negotiation domains and against 12 representative opponents demonstrate the effectiveness of our BDI framework.

Minh Sao Khue Luu, Bair N. Tuchinov

We introduce a single–backbone foundation model for brain MRI that supports dynamic modality integration: it operates with arbitrary, possibly unseen, combinations of MRI sequences at pretrain and transfer. The encoder is conditioned by text-derived modality embeddings via conditional layer normalization, while a variance–covariance penalty discourages feature collapse. Unlike expert-based designs that grow with each new sequence, our approach scales without adding modality-specific branches. Pretrained self-supervised on ∼60,000 heterogeneous MRIs, the model learns modality-aware yet modality-agnostic features. We outline evaluation on segmentation and classification under missing/unseen modalities and cross-center shifts, and present early feasibility on multiple sclerosis lesion segmentation under limited data. This work moves toward robust, protocol-agnostic MRI foundation models suited to real clinical variability.

Xinyan Liu, Zhuoning Xu

Jigsaw puzzle solving remains difficult because models must reconcile local fragment cues with global structure. Most prior work leans solely on visual signals (edge or texture coherence) and rarely exploits natural-language descriptions, which are especially helpful for puzzles with eroded gaps. We introduce a vision–language framework that uses textual context to guide assembly. At its core, the Vision–Language Hierarchical Semantic Alignment (VLHSA) module aligns image patches with text via multi-level matching—from local tokens to global summaries—within a multimodal design that couples dual visual encoders with language features for cross-modal reasoning. Across multiple datasets, the method surpasses the state of the art, including a 14.2 percentage point gain in piece accuracy; ablations identify VLHSA as the principal source of improvement. These results suggest a practical shift for jigsaw solving: augmenting vision with language to resolve ambiguous placements

Xiaotao Liu, Prashant Mohit, Arvind Easwaran

In safe reinforcement learning (SRL), there exists an inherent conflict between maximizing reward and minimizing cost. We propose a novel approach that effectively resolve the conflict between maximizing reward and minimizing cost in joint optimization.When the cost exceeds the threshold, we perform cost-reducing updates. Otherwise, we compute policy gradients that maximize expected rewards, while using second-order Taylor approximation to evaluate whether these reward-maximizing gradients would violate the cost constraint. If constraint violation is detected, we adjust the gradient direction to maintain safety compliance; otherwise, we execute standard reward-increasing policy updates. This approach helps ensure that reward-seeking updates do not inadvertently increase costs, thereby reducing the likelihood of constraint violations. Empirical tests show our framework successfully manages reward-cost trade-offs through reward augmentation and cost shaping, improving both performance and safety without switching optimization strategies. Results demonstrate that concurrent treatment of both objectives in one policy gradient update is viable for improving safe reinforcement learning methods.

Pengfei Liu, Rui Kong, Zongzhang Zhang

Hand-craft reward engineering requires domain knowledge with numerous trials and errors, while Preference-based Reinforcement Learning (PbRL) avoids manual reward design but often suffers from limited interpretability and unstable training. To address these issues, we propose a novel preference alignment framework. Our approach leverages large language models to generate sub-reward functions informed by prior knowledge and further align human preferences by optimizing the weights combining these sub-rewards. For policy learning, we introduce Policy Optimization via Pareto Regularization (POPR) which regularizes updates along Pareto-optimal directions. Experiments show that our framework improves reward quality and policy stability, achieving superior performance to expert-designed rewards across most tasks.

Lianghui Liu, Zongzhang Zhang

Offline Meta-Reinforcement Learning (OMRL) leverages pre-collected data to adapt to new tasks. Context-based methods learn task representations from contexts. However, the context is influenced by both the task and the behavior policy. The mismatch between the behavior policy and the testing policy causes a context distribution shift problem, which results in poor task representations and degraded performance. This problem is exacerbated in settings with data limitations. To address this, we propose a novel approach called Meta-Normalizing Flow (Meta-NF). First, it employs a highly expressive and sample-efficient normalizing flow policy. Second, it incorporates a metric for testing-time task representation selection to effectively mitigate the context shift problem. Empirical results demonstrate that Meta-NF outperforms existing OMRL methods, with both components contributing to its strong performance.

Jizhihui Liu, Guangdao Zhu, Feiyi Du

Vision-Language Models (VLMs) encode images into lengthy sequences of visual tokens, leading to excessive computational overhead and limited inference efficiency. In this paper, we study the hierarchical attention pattern in vision encoders and propose HiPrune, a training-free and model-agnostic token Pruning framework for VLMs. We identify that middle layers in the vision encoder attend to object-centric regions, while deep layers capture global contextual features. Based on this observation, HiPrune selects tokens based on the attention score from the middle and deep layers. Our method requires no retraining and integrates seamlessly with any ViT-based VLM. Experiments demonstrate that HiPrune achieves outstanding pruning performance, maintaining a balance between efficiency and efficacy.

Yow-Fu Liou, Yu-Chien Tang, An-Zi Yen

When evaluating large language models (LLMs) for question answering tasks, a common protocol is multiple-choice question-answering (MCQA), where the model selects from a fixed set of choices. In contemporary robustness testing, researchers typically perturb instructions or introduce confusion into factual statements; however, model behavior also hinges on choice compliance: whether models remain within the canonical set A-D. We formalize this setting by asking whether the model continues to respect the interface's rules when the problem presents a tempting alternative. Our approach is interface-preserving: we append a single selectable option E while keeping the question and A-D unchanged. Then, we introduce three types of malicious option injection to assess LLMs' robustness. Experimental results highlight the vulnerability of LLMs on contradict type content of the additional option E. Our evaluation framework can effectively serve as a low-cost audit of rule adherence on existing datasets and black-box models, surfaces off-policy items, and supports interpretable model comparison for deployment.

Kunrong Li, Zhu Sun, Kwan Hui Lim

We propose HyMoERec, a novel sequential recommendation framework that addresses the limitations of uniform Position-wise Feed-Forward Networks in existing models. Current approaches treat all user interactions and items equally, overlooking the heterogeneity in user behavior patterns and diversity in item complexity. HyMoERec initially introduces a hybrid mixture-of-experts architecture that combines shared and specialized expert branches with an adaptive expert fusion mechanism for the sequential recommendation task. This design captures diverse reasoning for varied users and items while ensuring stable training. Experiments on MovieLens-1M and Beauty datasets demonstrate that HyMoERec consistently outperforms state-of-the-art baselines.

Jingcheng Li, Ye Qiao, Sitao Huang

Current video understanding models struggle with temporal reasoning and efficient processing while balancing detail preservation with computational efficiency. We propose a hierarchical memory system that segments videos into action and scene units, combined with question-aware agentic keyframe selection. Our method achieves 70.3% overall accuracy on VideoMME short video benchmarks.

Yousung Lee, Dongsoo Har

Dynamic head pruning in Vision Transformers (ViTs) improves efficiency by removing redundant attention heads, but existing pruning policies are often difficult to interpret and control. In this work, we propose a novel framework by integrating Sparse Autoencoders (SAEs) with dynamic pruning, leveraging their ability to disentangle dense embeddings into interpretable and controllable sparse latents. Specifically, we train an SAE on the final-layer residual embedding of the ViT and amplify the sparse latents with different strategies to alter pruning decisions. Among them, per-class steering reveals compact, class-specific head subsets that preserve accuracy. For example, bowl improves accuracy (76%→82%) while reducing head usage (0.72→0.33) via heads h2 and h5. These results show that sparse latent features enable class-specific control of dynamic pruning, effectively bridging pruning efficiency and mechanistic interpretability in ViTs.

Younsung Lee, Yujin Kim, Kyungjoong Kim

Esports is growing rapidly, yet the data available to researchers is limited due to the game company policies. Consequently, vision-based approaches utilizing game screens are gaining attention as a practical alternative. We focus on the League of Legends minimap and address the challenges of champion detection when extracting champion information from the minimap. The challenges in this domain include small objects, rapid movement, and frequent occlusions. We propose a transfer-learning-based object detection pipeline that combines synthetic data with a subset of replay data. Synthetic data enables the rapid generation of diverse scenarios and improves training scalability, while replay data reduces the data distribution gap. This approach achieves 0.588 mean average precision, improving over replay-only by 0.261 and synthetic-only by 0.312, with 6.4 ms latency. Furthermore, we constructed a dataset encompassing all champions, enabling comparative analysis of detection models and supporting reproducible benchmarking for various application studies.

Tae Hoon Lee, Joongheon Kim

In adaptive cruise control (ACC), balancing safety, comfort, and sustainability still remains challenging. Accordingly, we propose a hybrid reinforcement learning framework combining proximal policy optimization (PPO) and deep Q-network (DQN) with a multi-objective reward for autonomous carbon-neutral eco-driving. Experimental results revealed the contrasts between eco and non-eco modes, underscoring how reward design shapes driving behaviors.

Suwoong Lee, Seungjae Lee, Yunho Jeon, Junmo Kim

We propose Tailored ViT Slimming (TVS), a budget-aware multi-dimensional pruning framework for Vision Transformers. TVS injects learnable masks into MHSA and MLP modules and applies adaptive non-convex sparsity regularization to achieve maximal utilization of parameters under strict module-wise budgets. In addition, by retaining scaled masks after pruning, TVS avoids abrupt accuracy drops and provides stable initialization for fine-tuning. On ImageNet-1k with DeiT-S and DeiT-B, TVS consistently outperforms prior ViT compression methods. This result empirically shows that the non-convex sparsity regularizer is effective not only in CNNs but also in ViTs.

Seung Kwon Lee, Seok Bin Son, Joongheon Kim, Hoh Peter In

Quantum machine learning (QML) has attracted growing interest for their ability to achieve superior performance with significantly fewer parameters. However, the high cost and scarcity of current hardware push inference to cloud-hosted quantum devices, creating a tension between verifiability and confidentiality. This work proposes a novel framework that converts quantum neural network operations into classical arithmetic circuits that faithfully approximate genuine quantum computations. By encrypting these circuits with zero-knowledge proofs, it ensures computational validity while concealing internal parameters. Experimental results show that our classical circuits achieve fidelity above 0.9996 and total variation distance below 1% compared to actual quantum computations, verifying the practicality of trustworthy and privacy-preserving quantum inference.

Vishesh Kumar, Ahana Chanda, Poulomi Bhattacharya, Akshay Agarwal

Automated mosquito species identification is critical for combating vector-borne diseases. We introduce Q-MoFusion, a novel hybrid quantum-classical framework that fuses deep features from pre-trained Audio Spectrogram Transformer (AST) and Whisper models using a Variational Quantum Circuit (VQC). Our approach significantly outperforms individual backbones and prior state-of-the-art benchmarks, demonstrating superior accuracy and robustness, particularly on imbalanced classes. Q-MoFusion demonstrates the potential of hybrid quantum computing to enhance bioacoustic surveillance for addressing critical public health challenges.

Daiki Kitashima, Ryota Higa, Katsuhide Fujita

Automated negotiation, a form of interaction among autonomous agents, plays a central role in multi-agent systems, yet the application of large language model (LLM) in this domain remains underexplored. An LLM can serve as a meta-strategist, adaptively selecting explicit strategies for execution by external strategic tools based on its capabilities. We propose a negotiation AI agent equipped with explicit strategic tools, including time-dependent and tit-for-tat negotiation strategies. Our results show that strategic tool enhanced negotiators achieve approximately 16% higher average utility compared with baseline, latest LLM negotiators.