IEEE transactions on artificial intelligence最新文献_第4页

How to Ask the AI: A User Perspective Survey for Large Language Model Prompting 如何询问AI：大型语言模型提示的用户视角调查

IEEE transactions on artificial intelligence Pub Date : 2026-10-01 Epub Date: 2026-06-04 DOI: 10.1109/TAI.2026.3700136

Yiqun Zhang;Yunfan Zhang;Mingjie Zhao;Sen Feng;Yiu-ming Cheung

{"title":"How to Ask the AI: A User Perspective Survey for Large Language Model Prompting","authors":"Yiqun Zhang;Yunfan Zhang;Mingjie Zhao;Sen Feng;Yiu-ming Cheung","doi":"10.1109/TAI.2026.3700136","DOIUrl":"https://doi.org/10.1109/TAI.2026.3700136","url":null,"abstract":"AI tools such as ChatGPT and DeepSeek, powered by large language models (LLMs), allow users to obtain instant and effective content responses simply by typing requests, such as “plan a three-day Vienna trip,” “solve the attached mathematical problem,” “draft an email to inquire review progress,” etc., which are also known as LLM prompts. Crafting clear and well-structured prompts leads to more appropriate LLM feedback, which effectively bridges human-LLM interaction. Although prompting appears accessible to nonexpert users, precisely organizing effective prompts is a highly systematic and skilful process, presenting potential challenges even for experienced users. This survey explores the principles, taxonomy, and organization of prompts from a user-centered perspective. Differing from the existing surveys that primarily focus on technical principles and application scenarios of LLMs, this article provides actionable guidelines for formulating effective LLM prompts across diverse real-world tasks and specifically contributes by: 1) developing an intuitive evaluation strategy for prompt efficacy; 2) providing prompting workflow demonstrations on representative applications; and 3) maintaining a dynamically updated open-source project to ensure the core takeaways remain up-to-date. These measures lower the threshold for users to correctly understand and craft prompts that align with evolving application scenarios.","PeriodicalId":73305,"journal":{"name":"IEEE transactions on artificial intelligence","volume":"7 10","pages":"5320-5337"},"PeriodicalIF":0.0,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11551561","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097728","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Memory-Augmented Architecture for Long-Term Context Handling in Large Language Models 大型语言模型中长期上下文处理的记忆增强体系结构

IEEE transactions on artificial intelligence Pub Date : 2026-10-01 Epub Date: 2026-06-09 DOI: 10.1109/TAI.2026.3701680

Haseeb Ullah Khan Shinwari;Muhammad Usama

引用次数: 0

Multiitem Regularization of Deep Latent Variable Model for Semisupervised Data Regression 半监督数据回归中深度潜变量模型的多项目正则化

IEEE transactions on artificial intelligence Pub Date : 2026-10-01 Epub Date: 2026-03-26 DOI: 10.1109/TAI.2026.3696259

Junhua Zheng;Zhiqiang Ge;Le Yao

{"title":"Multiitem Regularization of Deep Latent Variable Model for Semisupervised Data Regression","authors":"Junhua Zheng;Zhiqiang Ge;Le Yao","doi":"10.1109/TAI.2026.3696259","DOIUrl":"https://doi.org/10.1109/TAI.2026.3696259","url":null,"abstract":"Deep latent variable models have recently been developed for industrial data analytics, with promising performances obtained from various application scenarios. However, there is a potential over-fitting risk with the increased model capacity, which may degrade their performances, particularly when there are only a limited number of training data samples. At the same time, a high-quality labeled dataset is much more difficult to obtain even we are in the age of big data, while the unlabeled dataset can be much more easily obtained. In this article, a new multiitem regularized deep latent variable regression modeling framework is proposed, with the introductions of both L2 norm and Laplacian regularization items. Therefore, the potential risk of over-fitting and the semisupervised learning problems can be unified in the same regularization modeling framework. Furthermore, basis and kernel functions are introduced to extend the linear semisupervised regression model to its nonlinear counterpart for possible performance enhancement. Two industrial examples are employed for performance evaluation and comparative studies among different methods.","PeriodicalId":73305,"journal":{"name":"IEEE transactions on artificial intelligence","volume":"7 10","pages":"5641-5650"},"PeriodicalIF":0.0,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097661","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Linearized Tensor Low-Rank Convolutional Network 线性化张量低秩卷积网络

IEEE transactions on artificial intelligence Pub Date : 2026-10-01 Epub Date: 2026-03-21 DOI: 10.1109/TAI.2026.3695504

Qiangqiang Shen;Zonglin Liu;Hanzhang Wang;Yongsheng Liang;Yongyong Chen;Shiqi Wang

{"title":"Linearized Tensor Low-Rank Convolutional Network","authors":"Qiangqiang Shen;Zonglin Liu;Hanzhang Wang;Yongsheng Liang;Yongyong Chen;Shiqi Wang","doi":"10.1109/TAI.2026.3695504","DOIUrl":"https://doi.org/10.1109/TAI.2026.3695504","url":null,"abstract":"To balance deep learning performance and interpretability, convolutional sparse coding (CSC) has successfully incorporated differentiable optimization layers in place of the standard convolutional layers of traditional deep neural networks (DNN). However, CSC models prefer to fall into local sparsity under the <inline-formula><tex-math>$boldsymbol{ell}_mathbf{1}$</tex-math></inline-formula>-norm, whose optimization shows limitations in preserving the geometric structure and global sparsity of tensor data. In this article, we propose a linearized tensor low-rank convolutional network (LTLRCNet) to refresh the differentiable optimization layers for DNN [e.g., Residual Network (ResNet) and U-shaped Network (UNet)], which instead optimizes the tensor low-rank minimization problem for global sparsity without ruining the intrinsic structure of data. Specifically, instead of simply replacing the <inline-formula><tex-math>$boldsymbol{ell}_mathbf{1}$</tex-math></inline-formula>-norm with a tensor low-rank regularizer, we first propose a linearized tensor low-rank representation (LTLRR) model optimized by a linearized alternating direction method to decouple the dictionary and code coefficient without introducing an auxiliary variable, whose convergence is proved theoretically. Subsequently, we extend the LTLRR model into a deep unfolding network, such as the LTLRCNet, by engineering a convolutional dictionary that functions as a differentiable optimization layer, effectively substituting the conventional convolutional layer. To match the deep unfolding framework and avoid the gradient explosion, we apply a rank-one tactic bearing the Frank-Wolfe algorithm to solve the tensor low-rank minimization problem instead of conducting the cumbersome full-rank tensor singular value decomposition. Experimental results on image classification and segmentation show the potential of our LTLRCNet with its modeling-based layers to maintain significant performance and interpretability.","PeriodicalId":73305,"journal":{"name":"IEEE transactions on artificial intelligence","volume":"7 10","pages":"5627-5640"},"PeriodicalIF":0.0,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097343","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Trident: Detecting Face Forgeries With Domain-Adversarial Triplet Learning 三叉戟：用领域对抗三联学习检测人脸伪造

IEEE transactions on artificial intelligence Pub Date : 2026-10-01 Epub Date: 2026-06-15 DOI: 10.1109/TAI.2026.3703428

Mustafa Hakan Kara;Aysegul Dundar;Uğur Güdükbay

{"title":"Trident: Detecting Face Forgeries With Domain-Adversarial Triplet Learning","authors":"Mustafa Hakan Kara;Aysegul Dundar;Uğur Güdükbay","doi":"10.1109/TAI.2026.3703428","DOIUrl":"https://doi.org/10.1109/TAI.2026.3703428","url":null,"abstract":"As face forgeries generated by deep neural networks become increasingly sophisticated, detecting face manipulations in digital media has posed a significant challenge, underscoring the importance of maintaining digital media integrity and combating visual disinformation. Current detection models, predominantly trained on domain-specific data, often falter against forgeries generated by unencountered techniques. In response to this challenge, we introduce <italic>Trident, a face forgery detection framework that employs triplet learning within a Siamese network architecture to enhance adaptability across diverse forgery methods. <italic>Trident is trained on curated triplets to isolate nuanced differences of forgeries, capturing fine-grained features that distinguish pristine samples from manipulated ones while controlling for other variables. To further enhance generalizability, we incorporate domain-adversarial training with a forgery discriminator. This adversarial component guides our embedding model toward forgery-agnostic representations, improving its robustness to unseen manipulations. In addition, we prevent gradient flow from the classifier head to the embedding model, thereby avoiding overfitting induced by artifacts specific to certain forgeries. Comprehensive evaluations across multiple benchmarks and ablation studies demonstrate the effectiveness of our framework.","PeriodicalId":73305,"journal":{"name":"IEEE transactions on artificial intelligence","volume":"7 10","pages":"5838-5851"},"PeriodicalIF":0.0,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11563481","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097316","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Robustness of Prompting: Enhancing Robustness of Large Language Models Against Prompt Attacks 提示的鲁棒性：增强大型语言模型对提示攻击的鲁棒性

IEEE transactions on artificial intelligence Pub Date : 2026-10-01 Epub Date: 2026-03-27 DOI: 10.1109/TAI.2026.3697684

Lin Mu;Guowei Chu;Li Ni;Lei Sang;Yiwen Zhang

{"title":"Robustness of Prompting: Enhancing Robustness of Large Language Models Against Prompt Attacks","authors":"Lin Mu;Guowei Chu;Li Ni;Lei Sang;Yiwen Zhang","doi":"10.1109/TAI.2026.3697684","DOIUrl":"https://doi.org/10.1109/TAI.2026.3697684","url":null,"abstract":"Large language models (LLMs) have demonstrated remarkable performance across various tasks by effectively utilizing a prompting strategy. However, they are highly sensitive to input perturbations, such as typographical errors or slight character order errors, which can significantly impair their performance. Despite advances in prompting techniques such as chain-of-thought (CoT) and automatic prompt generation, developing a prompting strategy that explicitly mitigates the negative impact of such perturbations remains an open challenge. To bridge this gap, we propose <italic>robustness <italic>of <italic>prompting (<italic>RoP), a novel prompting strategy aimed at enhancing the robustness of LLMs. RoP consists of two stages: <italic>Error Correction and <italic>Guidance. In the <italic>Error Correction stage, RoP applies diverse perturbation methods to generate adversarial examples, which are used to generate prompts that correct input errors automatically. In the <italic>Guidance stage, RoP generates an optimal guidance prompt based on the corrected input, guiding the model to generate more robust and accurate inferences. Through comprehensive experiments spanning arithmetic, commonsense, and logical reasoning tasks, we demonstrate that RoP significantly improves LLMs’ robustness against adversarial perturbations. Crucially, it preserves model accuracy with only minimal degradation compared to clean input scenarios, thereby establishing RoP as a practical and effective approach for enhancing LLM robustness in real-world applications.","PeriodicalId":73305,"journal":{"name":"IEEE transactions on artificial intelligence","volume":"7 10","pages":"5664-5677"},"PeriodicalIF":0.0,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097563","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Artifact-Robust Graph-Based Learning in Digital Pathology 数字病理学中基于人工鲁棒图的学习

IEEE transactions on artificial intelligence Pub Date : 2026-10-01 Epub Date: 2026-04-20 DOI: 10.1109/TAI.2026.3685637

Saba Heidari Gheshlaghi;Milan Aryal;Nasim Yahyasoltani;Masoud Ganji

{"title":"Artifact-Robust Graph-Based Learning in Digital Pathology","authors":"Saba Heidari Gheshlaghi;Milan Aryal;Nasim Yahyasoltani;Masoud Ganji","doi":"10.1109/TAI.2026.3685637","DOIUrl":"https://doi.org/10.1109/TAI.2026.3685637","url":null,"abstract":"Whole slide images (WSIs) are digitized images of tissues placed on glass slides using advanced scanners. The digital processing of WSIs is challenging as they are gigapixel images and stored in multiresolution format. A common challenge with WSIs is that perturbations/artifacts are inevitable during the storage of the glass slides and digitizing them. These perturbations include motion, which often arises from slide movement during placement, and changes in hue and brightness due to variations in staining chemicals and the quality of digitizing scanners. In this work, a novel robust learning approach is presented to account for these artifacts. Due to the size and resolution of WSIs and to account for neighborhood information, graph-based methods are called for. We use graph convolutional network (GCN) to extract features from the graph that represents the WSI. Using a denoiser and pooling layer, perturbations in WSIs are effectively managed. Subsequently, the output is processed by a transformer that classifies two different datasets: one for various grades of prostate cancer and another for lung cancer classification. To compare the efficacy of the proposed approach, the model without denoiser is trained and tested with WSIs without any perturbation and then different perturbations are introduced in WSIs and passed through the network with the denoiser. The accuracy and kappa scores of the proposed model with both datasets compared with non-robust algorithms show significant improvement in cancer diagnosis.","PeriodicalId":73305,"journal":{"name":"IEEE transactions on artificial intelligence","volume":"7 10","pages":"5383-5392"},"PeriodicalIF":0.0,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097559","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Ensemble of Transfer Techniques via Adaptive Resource Allocation for Multiobjective Multitask Optimization 基于自适应资源分配的多目标多任务优化集成迁移技术

IEEE transactions on artificial intelligence Pub Date : 2026-10-01 Epub Date: 2026-04-30 DOI: 10.1109/TAI.2026.3689513

Tingyu Zhang;Kuihua Huang;Xinyi Wu;Yanchi Li;Jianfeng Liu;Shuijia Li;Wenyin Gong

{"title":"Ensemble of Transfer Techniques via Adaptive Resource Allocation for Multiobjective Multitask Optimization","authors":"Tingyu Zhang;Kuihua Huang;Xinyi Wu;Yanchi Li;Jianfeng Liu;Shuijia Li;Wenyin Gong","doi":"10.1109/TAI.2026.3689513","DOIUrl":"https://doi.org/10.1109/TAI.2026.3689513","url":null,"abstract":"Multiobjective multitask optimization (MMTO) has obtained increasing attention recently, where effective knowledge transfer (KT) is essential for improving optimization effectiveness. Existing KT techniques primarily focus on transferring knowledge derived from individuals or overall population distributions, but have not fully exploited manifold-based information. The effectiveness of using manifold information to guide the search has been demonstrated in single-task multiobjective optimization. To address these challenges, we propose an ensemble-based transfer adaptation framework, named MMTO-ensemble-based transfer adaptation (ETA), which integrates multiple transfer techniques via dynamic resource allocation. Specifically, MMTO-ETA estimates the manifold of the current population for each task and extracts two types of knowledge: structure knowledge (SK) and location knowledge (LK), and incorporates corresponding transfer techniques. The SK-based and LK-based transfer techniques are inherently complementary, each leveraging its distinct strengths to enhance the transfer process. MMTO-ETA continuously evaluates online performance and allocates computational resources among self-evolution, SK-based transfer, and LK-based transfer. This ensures that the most effective approach is prioritized at each iteration, reducing the risk of negative transfer. Extensive experiments on two benchmark suites and a real-world application demonstrate that MMTO-ETA consistently outperforms eleven state-of-the-art algorithms.","PeriodicalId":73305,"journal":{"name":"IEEE transactions on artificial intelligence","volume":"7 10","pages":"5573-5586"},"PeriodicalIF":0.0,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097630","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A Unified Framework for Reliable Survival Analysis 可靠生存分析的统一框架

IEEE transactions on artificial intelligence Pub Date : 2026-10-01 Epub Date: 2026-03-27 DOI: 10.1109/TAI.2026.3697686

Armand Bandiang Massoua;Mohamed Bouguessa;Abdoulaye Baniré Diallo

{"title":"A Unified Framework for Reliable Survival Analysis","authors":"Armand Bandiang Massoua;Mohamed Bouguessa;Abdoulaye Baniré Diallo","doi":"10.1109/TAI.2026.3697686","DOIUrl":"https://doi.org/10.1109/TAI.2026.3697686","url":null,"abstract":"Survival analysis (time-to-event prediction) focuses on predicting the time until an event of interest, such as disease relapse or death. It is a critical tool for clinical decision-making. An effective survival approach must address model understandability, quantifying uncertainty, and predictive robustness for practical adoption and trustworthiness in clinical settings. While progress has been made in these areas, existing methods often address these aspects separately, leading to limited performance and practical challenges. In this article, we propose a principled framework that addresses these limitations in a unified approach. Our framework incorporates a stochastic, biologically inspired sparsity mechanism for feature selection and applies a Bayesian treatment of model weights, enabling double stochasticity through two distinct sampling processes. These mechanisms enhance model understandability and enable robust uncertainty quantification. Furthermore, we introduce an information-bound regularization inspired by information theory, which controls information flow by promoting the learning of essential information while filtering out irrelevant and redundant data. This regularization improves generalization and enhances predictive resilience to noise and adversarial perturbations. Through extensive experiments on five real-world survival datasets, our approach demonstrates competitive performance in both discrimination and calibration compared to state-of-the-art deep survival models. On a synthetic dataset, our approach effectively identifies relevant features, showcasing its ability to support reliable feature selection. Additionally, the robustness of the proposed framework to adversarial attacks highlights its predictive resilience, further reinforcing its trustworthiness for survival analysis applications.","PeriodicalId":73305,"journal":{"name":"IEEE transactions on artificial intelligence","volume":"7 10","pages":"5678-5692"},"PeriodicalIF":0.0,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097978","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

CTRL: Cross-Modal Transformer With Reinforcement Learning for Climate Causal Discovery CTRL：用于气候因果发现的跨模态变压器强化学习

IEEE transactions on artificial intelligence Pub Date : 2026-10-01 Epub Date: 2026-03-18 DOI: 10.1109/TAI.2026.3688211

Jinduo Liu;Ting Wang;Yilin Lu;Junzhong Ji

{"title":"CTRL: Cross-Modal Transformer With Reinforcement Learning for Climate Causal Discovery","authors":"Jinduo Liu;Ting Wang;Yilin Lu;Junzhong Ji","doi":"10.1109/TAI.2026.3688211","DOIUrl":"https://doi.org/10.1109/TAI.2026.3688211","url":null,"abstract":"Understanding the complex internal relationships of climate elements and revealing causal relationships in climate phenomena has attracted attention in the field of climate science and data mining. However, current causal discovery methods in climate science primarily rely on single-modal data, which limits the available information and complicates the accurate identification of climate causal relationships. In this article, we propose a novel climate causal discovery method based on cross-modal Transformer and reinforcement learning (RL), called CTRL. The method consists of three modules: unimodal feature representation, cross-modal feature fusion, and RL for causal discovery. Specifically, we first design a novel unimodal feature representation method that uses the seasonal-trend decomposition based on LOESS (STL) strategy to decompose air quality and meteorological features into trend and seasonal components, with the aim of effectively capturing temporal data patterns. Next, we develop a cross-modal feature fusion method using cross-attention to capture cross-modal interactions and adaptively fuse bimodal features. Finally, we employ RL to search for the optimal climate causal network with the best Bayesian information criterion (BIC) score. Numerous experiments show that CTRL outperforms single-modal approaches in climate causal discovery, achieving even better results when predicting a target climate variable using only the causal features identified by CTRL.","PeriodicalId":73305,"journal":{"name":"IEEE transactions on artificial intelligence","volume":"7 10","pages":"5524-5536"},"PeriodicalIF":0.0,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149097230","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0