Pub Date : 2025-09-15DOI: 10.1109/MSP.2025.3572636
Shahana Ibrahim;Panagiotis A. Traganitis;Xiao Fu;Georgios B. Giannakis
One of the primary catalysts fueling advances in artificial intelligence (AI) and machine learning (ML) is the availability of massive, curated datasets. A commonly used technique to curate such massive datasets is crowdsourcing, where data are dispatched to multiple annotators. The annotatorproduced labels are then fused to serve downstream learning and inference tasks. This annotation process often creates noisy labels due to various reasons, such as the limited expertise, or unreliability of annotators, among others. Therefore, a core objective in crowdsourcing is to develop methods that effectively mitigate the negative impact of such label noise on learning tasks. This feature article introduces advances in learning from noisy crowdsourced labels. The focus is on key crowdsourcing models and their methodological treatments, from classical statistical models to recent deep learningbased approaches, emphasizing analytical insights and algorithmic developments. In particular, this article reviews the connections between signal processing (SP) theory and methods, such as identifiability of tensor and nonnegative matrix factorization, and novel, principled solutions of longstanding challenges in crowdsourcing—showing how SP perspectives drive the advancements of this field. Furthermore, this article touches upon emerging topics that are critical for developing cutting-edge AI/ML systems, such as crowdsourcing in reinforcement learning with human feedback (RLHF) and direct preference optimization (DPO) that are key techniques for fine-tuning large language models (LLMs).
{"title":"Learning From Crowdsourced Noisy Labels: A signal processing perspective","authors":"Shahana Ibrahim;Panagiotis A. Traganitis;Xiao Fu;Georgios B. Giannakis","doi":"10.1109/MSP.2025.3572636","DOIUrl":"https://doi.org/10.1109/MSP.2025.3572636","url":null,"abstract":"One of the primary catalysts fueling advances in <italic>artificial intelligence</i> (AI) and <italic>machine learning</i> (ML) is the availability of massive, curated datasets. A commonly used technique to curate such massive datasets is crowdsourcing, where data are dispatched to multiple annotators. The annotatorproduced labels are then fused to serve downstream learning and inference tasks. This annotation process often creates noisy labels due to various reasons, such as the limited expertise, or unreliability of annotators, among others. Therefore, a core objective in crowdsourcing is to develop methods that effectively mitigate the negative impact of such label noise on learning tasks. This feature article introduces advances in learning from noisy crowdsourced labels. The focus is on key crowdsourcing models and their methodological treatments, from classical statistical models to recent deep learningbased approaches, emphasizing analytical insights and algorithmic developments. In particular, this article reviews the connections between signal processing (SP) theory and methods, such as identifiability of tensor and nonnegative matrix factorization, and novel, principled solutions of longstanding challenges in crowdsourcing—showing how SP perspectives drive the advancements of this field. Furthermore, this article touches upon emerging topics that are critical for developing cutting-edge AI/ML systems, such as crowdsourcing in reinforcement learning with human feedback (RLHF) and direct preference optimization (DPO) that are key techniques for fine-tuning large language models (LLMs).","PeriodicalId":13246,"journal":{"name":"IEEE Signal Processing Magazine","volume":"42 3","pages":"84-106"},"PeriodicalIF":9.6,"publicationDate":"2025-09-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145061912","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-03-28DOI: 10.1109/MSP.2025.3569395
Nir Shlezinger;Guy Revach;Anubhab Ghosh;Saikat Chatterjee;Shuo Tang;Tales Imbiriba;Jindrich Dunik;Ondrej Straka;Pau Closas;Yonina C. Eldar
The Kalman filter (KF) and its variants are among the most celebrated algorithms in signal processing. These methods are used for state estimation of dynamic systems by relying on mathematical representations in the form of simple state-space (SS) models, which may be crude and inaccurate descriptions of the underlying dynamics. Emerging data-centric artificial intelligence (AI) techniques tackle these tasks using deep neural networks (DNNs), which are model agnostic. Recent developments illustrate the possibility of fusing DNNs with classic Kalman-type filtering, obtaining systems that learn to track in partially known dynamics. This article provides a tutorial-style overview of design approaches for incorporating AI in aiding KF-type algorithms. We review both generic and dedicated DNN architectures suitable for state estimation and provide a systematic presentation of techniques for fusing AI tools with KFs and for leveraging partial SS modeling and data, categorizing design approaches into task oriented and SS model oriented. The usefulness of each approach in preserving the individual strengths of model-based KFs and data-driven DNNs is investigated in a qualitative and quantitative study (whose code is publicly available), illustrating the gains of hybrid model-based/data-driven designs. We also discuss existing challenges and future research directions that arise from fusing AI and Kalman-type algorithms.
{"title":"Artificial Intelligence-Aided Kalman Filters: AI-Augmented Designs for Kalman-Type Algorithms","authors":"Nir Shlezinger;Guy Revach;Anubhab Ghosh;Saikat Chatterjee;Shuo Tang;Tales Imbiriba;Jindrich Dunik;Ondrej Straka;Pau Closas;Yonina C. Eldar","doi":"10.1109/MSP.2025.3569395","DOIUrl":"10.1109/MSP.2025.3569395","url":null,"abstract":"The Kalman filter (KF) and its variants are among the most celebrated algorithms in signal processing. These methods are used for state estimation of dynamic systems by relying on mathematical representations in the form of simple state-space (SS) models, which may be crude and inaccurate descriptions of the underlying dynamics. Emerging data-centric artificial intelligence (AI) techniques tackle these tasks using deep neural networks (DNNs), which are model agnostic. Recent developments illustrate the possibility of fusing DNNs with classic Kalman-type filtering, obtaining systems that learn to track in partially known dynamics. This article provides a tutorial-style overview of design approaches for incorporating AI in aiding KF-type algorithms. We review both generic and dedicated DNN architectures suitable for state estimation and provide a systematic presentation of techniques for fusing AI tools with KFs and for leveraging partial SS modeling and data, categorizing design approaches into task oriented and SS model oriented. The usefulness of each approach in preserving the individual strengths of model-based KFs and data-driven DNNs is investigated in a qualitative and quantitative study (whose code is publicly available), illustrating the gains of hybrid model-based/data-driven designs. We also discuss existing challenges and future research directions that arise from fusing AI and Kalman-type algorithms.","PeriodicalId":13246,"journal":{"name":"IEEE Signal Processing Magazine","volume":"42 3","pages":"52-76"},"PeriodicalIF":9.6,"publicationDate":"2025-03-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144164863","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Near-field (NF) signal processing introduces a new epoch in communication and sensing realms, showcasing transformative potential, particularly in extremely large-scale (XL) aperture array (ELAA) systems compared to its far-field (FF) counterpart. The NF spherical wavefront, incorporating the distance/range parameter through amplitude variations and phase differences among antennas, enhances spatial sensing capabilities. Localization, often intertwined with angle estimation, emerges as a direct beneficiary of this phenomenon, commanding substantial research attention. Moreover, the NF effects on spatial channels in ELAA communications mandate the formulation of diverse NF channel estimation (CE) methods. In this vein, our study presents a tutorial review of NF CE and localization, encapsulating fundamental wavefront models and extended advanced scenarios. Recognizing their pivotal roles in integrated sensing and communication (ISAC) systems, we examine their similarities and explore NF-integrated CE and localization (NF-ICEL) at the signal processing level. Additionally, we analyze system-level NF-ICEL under three specific scenarios, comparing them with FF-ICEL and highlighting the unique abilities and potential uses of NF-ICEL in scatterer/environment sensing, high-mobility situations, and unsynchronized systems.
{"title":"Near-Field Channel Estimation and Localization: Recent developments, cooperative integration, and future directions","authors":"Songjie Yang;Hua Chen;Wei Liu;Xiao-Ping Zhang;Chau Yuen","doi":"10.1109/MSP.2024.3500791","DOIUrl":"https://doi.org/10.1109/MSP.2024.3500791","url":null,"abstract":"Near-field (NF) signal processing introduces a new epoch in communication and sensing realms, showcasing transformative potential, particularly in extremely large-scale (XL) aperture array (ELAA) systems compared to its far-field (FF) counterpart. The NF spherical wavefront, incorporating the distance/range parameter through amplitude variations and phase differences among antennas, enhances spatial sensing capabilities. Localization, often intertwined with angle estimation, emerges as a direct beneficiary of this phenomenon, commanding substantial research attention. Moreover, the NF effects on spatial channels in ELAA communications mandate the formulation of diverse NF channel estimation (CE) methods. In this vein, our study presents a tutorial review of NF CE and localization, encapsulating fundamental wavefront models and extended advanced scenarios. Recognizing their pivotal roles in integrated sensing and communication (ISAC) systems, we examine their similarities and explore NF-integrated CE and localization (NF-ICEL) at the signal processing level. Additionally, we analyze system-level NF-ICEL under three specific scenarios, comparing them with FF-ICEL and highlighting the unique abilities and potential uses of NF-ICEL in scatterer/environment sensing, high-mobility situations, and unsynchronized systems.","PeriodicalId":13246,"journal":{"name":"IEEE Signal Processing Magazine","volume":"42 1","pages":"60-73"},"PeriodicalIF":9.4,"publicationDate":"2025-03-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143667329","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-03-21DOI: 10.1109/MSP.2024.3486470
Amir Masoud Molaei;Shaoqing Hu;Vincent Fusco;Thomas Fromenteze;Rupesh Kumar;Muhammad Ali Babar Abbasi;Okan Yurduseven
Near-field (NF) microwave and millimeter-wave (mm-wave) imaging, extending into the terahertz (THz) frequency range, has seen remarkable advancements across diverse applications, particularly security screening. These technologies benefit from the unique properties of microwave, mm-wave, and THz (MMT) spectra, such as penetration, nonionizing radiation, material sensitivity and the capability to operate in all weather conditions. This article provides an overview of the evolution and current state of NF radar imaging, emphasizing the critical role of signal processing in overcoming challenges related to hardware complexity, long acquisition time, and image reconstruction quality. Advanced signal processing techniques—including Fourier-based algorithms, sparse imaging, low-rank matrix recovery, and deep learning—are highlighted for their contributions to enhancing image resolution and processing efficiency. The article also discusses recent innovations in antenna technologies, aperture configurations, and scanning methods that have significantly improved NF radar imaging capabilities. Future research directions are suggested to further advance the field, highlighting the importance of continued exploration and innovation in NF MMT imaging.
{"title":"Advanced Near-Field Radar Imaging Approaches in Security: An overview on signal processing challenges, opportunities, and future directions","authors":"Amir Masoud Molaei;Shaoqing Hu;Vincent Fusco;Thomas Fromenteze;Rupesh Kumar;Muhammad Ali Babar Abbasi;Okan Yurduseven","doi":"10.1109/MSP.2024.3486470","DOIUrl":"https://doi.org/10.1109/MSP.2024.3486470","url":null,"abstract":"Near-field (NF) microwave and millimeter-wave (mm-wave) imaging, extending into the terahertz (THz) frequency range, has seen remarkable advancements across diverse applications, particularly security screening. These technologies benefit from the unique properties of microwave, mm-wave, and THz (MMT) spectra, such as penetration, nonionizing radiation, material sensitivity and the capability to operate in all weather conditions. This article provides an overview of the evolution and current state of NF radar imaging, emphasizing the critical role of signal processing in overcoming challenges related to hardware complexity, long acquisition time, and image reconstruction quality. Advanced signal processing techniques—including Fourier-based algorithms, sparse imaging, low-rank matrix recovery, and deep learning—are highlighted for their contributions to enhancing image resolution and processing efficiency. The article also discusses recent innovations in antenna technologies, aperture configurations, and scanning methods that have significantly improved NF radar imaging capabilities. Future research directions are suggested to further advance the field, highlighting the importance of continued exploration and innovation in NF MMT imaging.","PeriodicalId":13246,"journal":{"name":"IEEE Signal Processing Magazine","volume":"42 1","pages":"126-146"},"PeriodicalIF":9.4,"publicationDate":"2025-03-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143667552","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：