Artificial Intelligence Review最新文献

{"title":"Policy weighting via discounted Thomson sampling for non-stationary market-making","authors":"Óscar Fernández Vicente,&nbsp;Javier García,&nbsp;Fernando Fernández","doi":"10.1007/s10462-025-11312-9","DOIUrl":"10.1007/s10462-025-11312-9","url":null,"abstract":"<div><p>Market-making is an essential activity in every financial market. They provide liquidity to the system by placing buy and sell orders at multiple price levels. While performing this task, they aim to earn profit and manage inventory levels simultaneously. However, financial markets are not stationary environments; they constantly evolve, influenced by changes in participants, the occurrence of economic events, or the market trading hours, among others. This study introduces a novel approach to address the challenge of market-making in non-stationary financial markets with multi-objective Reinforcement Learning (RL). Traditional RL methods often struggle when applied to non-stationary environments, as the learned optimal policy may not be adapted to the new dynamics. We present Policy Weighting through Discounted Thompson Sampling (POW-dTS), a novel dynamic algorithm that adapts to changing market conditions by effectively weighting pre-trained policies across various contexts. Unlike some conventional methods, POW-dTS does not require additional artifacts such as change-point detection or models of transitions, making it robust against the unpredictability inherent in financial markets. Our approach focuses on optimizing trade profitability and managing inventory risk, the dual objectives of market makers. Through a detailed comparative analysis, we highlight the strengths and adaptability of POW-dTS against traditional techniques in non-stationary environments, demonstrating its potential to enhance market liquidity and efficiency.</p></div>","PeriodicalId":8449,"journal":{"name":"Artificial Intelligence Review","volume":"58 10","pages":""},"PeriodicalIF":13.9,"publicationDate":"2025-07-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1007/s10462-025-11312-9.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145144217","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Content moderation by LLM: from accuracy to legitimacy","authors":"Tao Huang","doi":"10.1007/s10462-025-11328-1","DOIUrl":"10.1007/s10462-025-11328-1","url":null,"abstract":"<div><p>One trending application of LLM (large language model) is to use it for content moderation in online platforms. Most current studies on this application have focused on the metric of <i>accuracy</i>—the extent to which LLMs make correct decisions about content. This article argues that accuracy is insufficient and misleading because it fails to grasp the distinction between easy cases and hard cases, as well as the inevitable trade-offs in achieving higher accuracy. Closer examination reveals that content moderation is a constitutive part of platform governance, the key to which is to gain and enhance <i>legitimacy</i>. Instead of making moderation decisions correctly, the chief goal of LLMs is to make them legitimate. In this regard, this article proposes a paradigm shift from the single benchmark of accuracy towards a legitimacy-based framework for evaluating the performance of LLM moderators. The framework suggests that for easy cases, the key is to ensure accuracy, speed, and transparency, while for hard cases, what matters is reasoned justification and user participation. Examined under this framework, LLMs’ real potential in moderation is not accuracy improvement. Rather, LLMs can better contribute in four other aspects: to conduct screening of hard cases from easy cases, to provide quality explanations for moderation decisions, to assist human reviewers in getting more contextual information, and to facilitate user participation in a more interactive way. To realize these contributions, this article proposes a workflow for incorporating LLMs into the content moderation system. Using normative theories from law and social sciences to critically assess the new technological application, this article seeks to redefine LLMs’ role in content moderation and redirect relevant research in this field.</p></div>","PeriodicalId":8449,"journal":{"name":"Artificial Intelligence Review","volume":"58 10","pages":""},"PeriodicalIF":13.9,"publicationDate":"2025-07-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1007/s10462-025-11328-1.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145144218","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A survey of NLP methods for oncology in the past decade with a focus on cancer registry applications","authors":"Isaac Hands,&nbsp;Ramakanth Kavuluru","doi":"10.1007/s10462-025-11316-5","DOIUrl":"10.1007/s10462-025-11316-5","url":null,"abstract":"<div><p>Clinical texts from pathology and radiology reports provide critical information for cancer diagnosis and staging. This study surveys the application of natural language processing (NLP) in cancer registry operations from 2014 to 2024. A total of 156 articles from Scopus and PubMed were reviewed and were categorized by NLP methods, document types, cancer sites, and research aims. NLP approaches were evenly distributed across rule-based (n=70), machine learning (n=66), and traditional deep learning (n=70), with transformer models (n=29) gaining prominence since 2019. Encoder-only models like BERT and its clinical adaptations (e.g., ClinicalBERT, RadBERT) show significant promise, though methods for increasing context length are needed. Decoder-only models (e.g., GPT-3, GPT-4) are less explored due to privacy concerns and computational demands. Notably, pediatric cancers, melanomas, and lymphomas are underrepresented, as are research areas such as disease progression, clinical trial matching, and patient communication. Multi-modal models, important for precision oncology and cancer screening, are also scarce. Our study highlights the potential of NLP to enhance data abstraction efficiency and accuracy in cancer registries, making greater use of cancer registry data for patient benefit. However, further research is needed to fully leverage transformer-based models, particularly for underrepresented cancer types and outcomes. Addressing these gaps can improve the timeliness, completeness, and accuracy of structured data collection from clinical text, ultimately enhancing cancer research and patient outcomes.</p></div>","PeriodicalId":8449,"journal":{"name":"Artificial Intelligence Review","volume":"58 10","pages":""},"PeriodicalIF":13.9,"publicationDate":"2025-07-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12267331/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144673817","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A review of named entity recognition: from learning methods to modelling paradigms and tasks","authors":"Wei Liang Seow,&nbsp;Iti Chaturvedi,&nbsp;Amber Hogarth,&nbsp;Rui Mao,&nbsp;Erik Cambria","doi":"10.1007/s10462-025-11321-8","DOIUrl":"10.1007/s10462-025-11321-8","url":null,"abstract":"<div><p>Named Entity Recognition (NER) is commonly used when summarising news articles and legal documents. It can extract the names of politicians or organisations and help determine the aspect of a positive or negative sentiment. Previous surveys have only provided a shallow review of NER with respect to a certain datatype. In contrast, here a much deeper coverage of different approaches is provided. First articles with respect to the learning method are discussed, such as supervised or unsupervised. Next, popular models that combine two or more learning methods are introduced in a bottom-up approach. The most popular NER algorithms are compared on a recently crawled 2024 election dataset from Australia. The effect of different parameters such as number of epochs and learning rate is explored. It is concluded that pre-trained NER models are limited in their ability to model new entities and disambiguate their context. Using the sentiment score together with a state space model over entities in a sentence might help overcome these challenges.</p></div>","PeriodicalId":8449,"journal":{"name":"Artificial Intelligence Review","volume":"58 10","pages":""},"PeriodicalIF":13.9,"publicationDate":"2025-07-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1007/s10462-025-11321-8.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145165665","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A recent advances on autism spectrum disorders in diagnosing based on machine learning and deep learning","authors":"Hajir Ammar Hatim,&nbsp;Zaid Abdi Alkareem Alyasseri,&nbsp;Norziana Jamil","doi":"10.1007/s10462-025-11302-x","DOIUrl":"10.1007/s10462-025-11302-x","url":null,"abstract":"<div><p>Neurological disorders affect communication ability, social interaction, and a person’s conduct. Early diagnosis and treatment of ASD during the early stages of a person’s life may result in better outcomes and a higher quality of life for patients. Current methods of diagnosis are based on behavioral observations and interviews, which are subjective, time-consuming, and costly. EEG does not include invasive techniques, and it is a safe and painless way of measuring electrical activity in the brain. EEG signals may reflect neural differences and abnormalities related to ASD and serve as a potential biomarker for diagnosis. Due to the increase in prevalence, there has been an increased need to develop more sensitive and unbiased diagnostic methods for ASD. ML and DL are two sophisticated methods that researchers developed for detecting ASD by doing neural network analyses. The review paper incorporates the analysis of previous studies; more than 200 works have been analyzed from top publishers like Elsevier, IEEE, MDPI, and Springer, specifically related to EEG signal analysis and feature extraction techniques. It considers significant methods for ASD detection, including SVMs, CNN, and other models like KNN, ResNet50, and ANFIS. Other datasets central in these studies are KAU, BCIAUT-P300, and ADOS-2. The performance metrics adopted in this research include accuracy, sensitivity, and specificity. For example, the cubic SVM realized an accuracy of 95.8%, while the CNN models reached 95%. Other models, like ResNet50, achieved 99.39%, while ANFIS reached 98.9%. Sensitivity and specificity also showed varying scores across the methods, between 85 and 100%, indicating the high potential of these approaches in ASD diagnosis. Future studies could pay more attention to dataset representativeness improvements and do the clinical validation of these models for better generalization and relevance toward early diagnosis in ASD.</p></div>","PeriodicalId":8449,"journal":{"name":"Artificial Intelligence Review","volume":"58 10","pages":""},"PeriodicalIF":13.9,"publicationDate":"2025-07-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1007/s10462-025-11302-x.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145166004","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A hybrid control approach to improve power quality in microgrid systems","authors":"Nima Khosravi","doi":"10.1007/s10462-025-11300-z","DOIUrl":"10.1007/s10462-025-11300-z","url":null,"abstract":"<div><p>Power quality (PQ) in distributed energy resources (DERs) is paramount for maintaining a stable and efficient electricity supply. The consistency and cleanliness of power are integral to ensuring reliability, sustainability, and optimal performance, thereby supporting a resilient and eco-friendly energy infrastructure. This paper introduces a hybrid control method designed to address two significant challenges in microgrid (MG) applications: active resonance damping (ARD) and unbalanced voltage compensation (UVC). Furthermore, the proposed hybrid method combines effective ARD with UVC at MG terminals. The active damping technique employs an external control level to counteract undesirable resonant harmonics, overcoming control bandwidth limitations. This approach offers simplicity in setup and performance without requiring additional system parameter adjustments. For UVC, the suggested control technique estimates the compensation reference using the dual d-q control, reducing the complexity and cost associated with load current measurement issues. The hybrid method integrates the resonant damping signal and the MG negative sequence reference (NSR) voltage, which are fed into a two-level sine-pulse width modulation block (SPWM) to control the MG converter. Simulation results validate the robustness of the proposed combined method in simultaneously compensating for unbalanced voltage and active resonance.</p><h3>Graphical abstract</h3><div><figure><div><div><picture><source><img></source></picture></div></div></figure></div></div>","PeriodicalId":8449,"journal":{"name":"Artificial Intelligence Review","volume":"58 10","pages":""},"PeriodicalIF":13.9,"publicationDate":"2025-07-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1007/s10462-025-11300-z.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145165918","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Benchmarking machine learning methods for the identification of mislabeled data","authors":"Lusine Nazaretyan,&nbsp;Ulf Leser,&nbsp;Martin Kircher","doi":"10.1007/s10462-025-11293-9","DOIUrl":"10.1007/s10462-025-11293-9","url":null,"abstract":"<div><p>Supervised machine learning recently gained growing importance in various fields of research. To train reliable models, data scientists need credible data, which is not always available. A particularly hard and widespread problem deteriorating the performance of methods are mislabeled samples (Northcutt in J Artif Intell Res 70:1373-1411, 2021). Common sources of mislabeling are weakly defined classes, labels that change their meaning, unsuitable annotators, or ambiguous guidelines for labeling. Because mislabeling lowers prediction quality, it is essential for scientists to be able to identify wrong labels before actually starting the learning process. For that, numerous algorithms for the identification of noisy instances have been developed. However, so far, a comprehensive empirical comparison of available methods has been missing.</p><p>In this paper, we survey and benchmark methods for the identification of mislabeled samples in tabular data. We discuss the theoretical background of label noise and how it can lead to mislabeling, review categorizations of identification methods, and briefly introduce 34 specific approaches together with popular data sets. Finally, 20 selected methods are benchmarked using artificially blurred data with controllable mislabeling and a new real-life genomic dataset with known errors. We compare methods varying the amount and the type of noise, as well as the sample size and domain of data. We find that most of the methods have the highest performance on datasets with a noise level of around 20-30% where the best filters identify around 80% of the noisy instances with relatively high precision (0.58<span>(-)</span>0.65). Acquiring precise predictions seems to be a more challenging task than identifying most of the noisy instances: while the average recall score over all models ranges from 0.48 to 0.77, the average precision score ranges from 0.16 to 0.55. Furthermore, none of the methods excels over all others in isolation, while ensemble-based methods often outperform individual models. We provide all data sets and analysis code to enable a better handling of mislabeled data and give recommendations on usage of noise filters depending on various dataset parameters.</p></div>","PeriodicalId":8449,"journal":{"name":"Artificial Intelligence Review","volume":"58 10","pages":""},"PeriodicalIF":13.9,"publicationDate":"2025-07-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1007/s10462-025-11293-9.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145165042","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Revolutionizing scholarly impact: advanced evaluations, predictive models, and future directions","authors":"Xiaomei Bai,&nbsp;Fuli Zhang,&nbsp;Jiaying Liu,&nbsp;Xiaoxia Wang,&nbsp;Feng Xia","doi":"10.1007/s10462-025-11315-6","DOIUrl":"10.1007/s10462-025-11315-6","url":null,"abstract":"<div><p>Artificial intelligence (AI) is revolutionising scholarly impact evaluation and prediction. By integrating AI and machine learning techniques, researchers can leverage diverse academic networks and multiple sources of academic big data. This integration transforms traditional evaluation methods that rely on structured measurements such as citation counts and journal impact factors, into more comprehensive and objective evaluations. In this paper, we dive deep into latest advancements in scholarly impact evaluation and prediction within the context of AI. We categorize existing models, highlighting their similarities and distinctions, with a particular emphasis on AI-enabled approaches. Building upon the analysis, we discuss the ongoing challenges in scholarly impact research and outline future directions in this field.</p></div>","PeriodicalId":8449,"journal":{"name":"Artificial Intelligence Review","volume":"58 10","pages":""},"PeriodicalIF":13.9,"publicationDate":"2025-07-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1007/s10462-025-11315-6.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145165376","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Fractional order dung beetle optimizer with reduction factor for global optimization and industrial engineering optimization problems","authors":"Huangzhi Xia,&nbsp;Yifen Ke,&nbsp;Riwei Liao,&nbsp;Huai Zhang","doi":"10.1007/s10462-025-11239-1","DOIUrl":"10.1007/s10462-025-11239-1","url":null,"abstract":"<div><p>Dung beetle optimizer (DBO) is a novel meta-heuristic algorithm inspired by the behaviors of dung beetles in nature, including ball rolling, dancing, foraging, stealing, and breeding. However, the standard DBO has weaknesses in global optimization, including the imbalance between the ability of exploration and exploitation, low accuracy in function solution, and susceptibility to falling into local optimum. To overcome the weaknesses of DBO, the fractional order dung beetle optimizer with reduction factor (FORDBO) is proposed. Firstly, the good nodes set sequence is employed to replace the randomly initialized population in the algorithm, aiming to enhance the diversity of the population. To enhance the global optimization performance of the algorithm, a reduction factor is designed to balance between the ability of exploration and exploitation. On the other hand, the fractional order calculus strategy is employed to adjust the dynamic boundary of the optimization region. The strategy enables the algorithm to focus on exploiting the potential optimization region. Finally, the repetitive renewal mechanism of the pathfinder dung beetle is proposed to enhance the ability of the algorithm to escape the local optimum. To evaluate the performance of FORDBO, on the one hand, we analyze the complexity of FORDBO and prove its convergence mathematically in this work. On the other hand, this work also compares the FORDBO with 23 similar swarm intelligence technologies through CEC2005, CEC2017, and CEC2022 benchmark functions for global optimization. At the same time, the FORDBO is applied to six industrial engineering optimization problems. The experimental numerical results show that the performance of FORDBO is better than other most swarm intelligence technologies. The source code of FORDBO is publicly available at https://github.com/Huangzhi-Xia/FORDBO.</p></div>","PeriodicalId":8449,"journal":{"name":"Artificial Intelligence Review","volume":"58 10","pages":""},"PeriodicalIF":13.9,"publicationDate":"2025-07-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1007/s10462-025-11239-1.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145164937","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Advances in artificial intelligence for olfaction and gustation: a comprehensive review","authors":"Zhihao Hao,&nbsp;Haisheng Li,&nbsp;Jianhua Guo,&nbsp;Yong Xu","doi":"10.1007/s10462-025-11309-4","DOIUrl":"10.1007/s10462-025-11309-4","url":null,"abstract":"<div><p>This review explores the transformative role of artificial intelligence (AI) in enhancing our understanding of olfaction and gustation, two senses that significantly influence human behavior and decision-making. AI methodologies, including machine learning and neural networks, are revolutionizing sensory experiences in industries such as food, fragrance, and healthcare. In the food industry, AI is enhancing flavor profiling, ensuring product safety, and aligning offerings with consumer preferences, all while preserving nutritional value. In fragrance, AI is enabling personalized scent creation, allowing for bespoke products tailored to diverse consumer needs. In healthcare, AI is advancing the diagnosis and treatment of sensory disorders, ultimately improving the quality of life for individuals with sensory impairments. Despite these advancements, challenges persist, such as the need for diverse and representative datasets, the subjective nature of sensory perception, and ethical concerns surrounding privacy. Addressing these issues requires interdisciplinary collaboration across neuroscience, computer science, and food technology. Future research should focus on developing adaptive AI systems that can dynamically interpret and predict sensory preferences, reshaping interactions with food, fragrances, and health products. This review underscores AI’s critical role in driving sensory innovation and lays the groundwork for future studies aimed at enhancing consumer experiences and improving quality of life.</p></div>","PeriodicalId":8449,"journal":{"name":"Artificial Intelligence Review","volume":"58 10","pages":""},"PeriodicalIF":13.9,"publicationDate":"2025-07-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1007/s10462-025-11309-4.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145164936","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}