IEEE Transactions on Emerging Topics in Computational Intelligence最新文献

{"title":"Equilibrium Contrastive Learning for Imbalanced Image Classification","authors":"Sumin Roh;Harim Kim;Ho Yun Lee;Il Yong Chun","doi":"10.1109/TETCI.2026.3670635","DOIUrl":"https://doi.org/10.1109/TETCI.2026.3670635","url":null,"abstract":"The data imbalance issue is critical in image classification. Contrastive learning (CL) is a predominant technique in image classification, but they showed limited classification performance when the class distribution in a training dataset is skewed (i.e., an imbalanced dataset). Recently, several supervised CL methods have been proposed particularly to promote an ideal regular simplex geometric configuration in the representation space—characterized by intra-class feature collapse and uniform inter-class mean spacing—especially for imbalanced datasets. In particular, existing prototype-based methods include class representatives, i.e., prototypes, as additional samples to achieve a more balanced treatment of all classes. However, the existing supervised CL methods for imbalanced datasets still suffer from two major limitations. First, they do <italic>not</i> consider the alignment between the class means/prototypes and linear classifiers, which could lead to poor generalization. Second, existing prototype-based methods treat class prototypes as <italic>only</i> one additional sample per class, making their influence depend on the number of class instances in a batch and causing unbalanced contributions across classes. To address these limitations, we propose Equilibrium Contrastive Learning (ECL), a supervised CL framework designed to promote geometric equilibrium in the representation space, where class features, means, and classifiers are harmoniously balanced under data imbalance. The proposed ECL framework uses two main strategies. First, ECL promotes the representation geometric equilibrium (i.e., a regular simplex geometry characterized by collapsed class samples and uniformly distributed class means), while balancing the contributions of class-average features and class prototypes. Second, ECL establishes a classifier-class center geometric equilibrium by aligning classifier weights and class prototypes in the representation space. We ran experiments with three long-tailed natural image datasets, the CIFAR-10-LT, CIFAR-100-LT and ImageNet-LT benchmark datasets, and the two imbalanced medical image classification datasets, the ISIC 2019 benchmark and our constructed lung cancer chest computed tomography dataset. Results with the five imbalanced datasets show that ECL outperforms existing state-of-the-art supervised CL methods designed for imbalanced/long-tailed classification. Codes are available at this link.","PeriodicalId":13135,"journal":{"name":"IEEE Transactions on Emerging Topics in Computational Intelligence","volume":"10 3","pages":"2429-2442"},"PeriodicalIF":6.0,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148515054","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A Two-Stage Learning Guided Multi-Objective Evolutionary Algorithm for Task Offloading in Edge Computing","authors":"Shijia Huang;Feng Wang;Jian Wang;Bing Li","doi":"10.1109/TETCI.2026.3670669","DOIUrl":"https://doi.org/10.1109/TETCI.2026.3670669","url":null,"abstract":"Task offloading in edge computing has become an effective way for reducing terminal device load and task execution time. Recently, various task offloading methods have been proposed in an effort to increase processing performance for computing tasks between edge servers and consumers. However, large-scale computing tasks need to be offloaded to edge nodes due to the terminal devices’ rapid development, which makes it difficult to obtain the optimal offloading scheme. Meanwhile, existing offloading methods usually consider a few common objectives and ignore other important factors such as economic cost. In this paper, the task offloading involving numerous computing tasks is formulated as a large-scale multi-objective optimization problem, which encompasses multiple optimization objectives, i.e., task time delay, economic cost and load balancing. Then, a two-stage learning guided multi-objective evolutionary algorithm (TSLEA) is proposed to generate task offloading schemes. Firstly, a global search based on the adaptive incremental histogram learning model is proposed to improve the exploration ability of TSLEA. Secondly, a local search based on decision space division is proposed to enhance the population convergence. Experimental results on three real-world datasets show that TSLEA achieves competitive performance compared to several representative algorithms.","PeriodicalId":13135,"journal":{"name":"IEEE Transactions on Emerging Topics in Computational Intelligence","volume":"10 3","pages":"2376-2390"},"PeriodicalIF":6.0,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148514417","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Automatic and Effective Discovery of Quantum Kernels","authors":"Massimiliano Incudini;Daniele Lizzio Bosco;Francesco Martini;Michele Grossi;Giuseppe Serra;Alessandra Di Pierro","doi":"10.1109/TETCI.2024.3499993","DOIUrl":"https://doi.org/10.1109/TETCI.2024.3499993","url":null,"abstract":"Quantum computing can empower machine learning models by enabling kernel machines to leverage quantum kernels for representing similarity measures between data. Quantum kernels are able to capture relationships in the data that are not efficiently computable on classical devices. However, there is no straightforward method to engineer the optimal quantum kernel for each specific use case. We present an approach to this problem, which employs optimization techniques, similar to those used in neural architecture search and AutoML, to automatically find an optimal kernel in a heuristic manner. To this purpose we define an algorithm for constructing a quantum circuit implementing the similarity measure as a combinatorial object, which is evaluated based on a cost function and then iteratively modified using a meta-heuristic optimization technique. The cost function can encode many criteria ensuring favorable statistical properties of the candidate solution, such as the rank of the Dynamical Lie Algebra. Importantly, our approach is independent of the optimization technique employed. The results obtained by testing our approach on a high-energy physics problem demonstrate that, in the best-case scenario, we can either match or improve testing accuracy with respect to the manual design approach, showing the potential of our technique to deliver superior results with reduced effort.","PeriodicalId":13135,"journal":{"name":"IEEE Transactions on Emerging Topics in Computational Intelligence","volume":"10 3","pages":"2210-2219"},"PeriodicalIF":6.0,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10812182","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148514522","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"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":"Learning Informative Latent Representation for Quantum State Tomography","authors":"Hailan Ma;Zhenhong Sun;Daoyi Dong;Dong Gong","doi":"10.1109/TETCI.2025.3543767","DOIUrl":"https://doi.org/10.1109/TETCI.2025.3543767","url":null,"abstract":"Quantum state tomography (QST) is the process of reconstructing the complete state of a quantum system (mathematically described as a density matrix) through a series of different measurements. These measurements are performed on a number of identical copies of the quantum system, with outcomes gathered as probabilities/frequencies. QST aims to recover the density matrix and the corresponding properties of the quantum state from the measured frequencies. Although an informationally complete set of measurements can specify the quantum state accurately in an ideal scenario with a large number of identical copies, both the measurements and identical copies are restricted and imperfect in practical scenarios, making QST highly ill-posed. The conventional QST methods usually assume adequate or accurate measured frequencies or rely on manually designed regularizers to handle the ill-posed reconstruction problem, suffering from limited applications in realistic scenarios. Recent advances in deep neural networks (DNNs) led to the emergence of deep learning (DL) in QST. However, existing DL-based QST approaches often employ generic DNN models that are not optimized for imperfect conditions of QST. In this paper, we propose a transformer-based autoencoder architecture tailored for QST with imperfect measurement data. Our method leverages a transformer-based encoder to extract <italic>an informative latent representation</i> (ILR) from imperfect measurement data and employs a decoder to predict the quantum states based on the ILR. We anticipate that the high-dimensional ILR will capture more comprehensive information about the quantum states. To achieve this, we conduct pre-training of the encoder using a pretext task that involves reconstructing high-quality frequencies from measured frequencies. Extensive simulations and experiments demonstrate the remarkable ability of the informative latent representation to deal with imperfect measurement data in QST.","PeriodicalId":13135,"journal":{"name":"IEEE Transactions on Emerging Topics in Computational Intelligence","volume":"10 3","pages":"2199-2209"},"PeriodicalIF":6.0,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148509177","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Non-Negative Gated Recurrent Neural Networks","authors":"Loukia Avramelou;Manos Kirtas;Nikolaos Passalis;Nikolaos Pleros;Anastasios Tefas","doi":"10.1109/TETCI.2026.3668523","DOIUrl":"https://doi.org/10.1109/TETCI.2026.3668523","url":null,"abstract":"Neural networks have become increasingly popular in recent years due to their ability to efficiently solve a wide range of complex problems, including computer vision, machine translation, and speech recognition. The core components of neural networks are their learnable parameters, which consist of their weights and biases. In typical neural networks, these parameters can be positive or negative, allowing models to learn complicated patterns in data. However, having both positive and negative values in a network introduces challenges in cases of analog hardware deployments. For instance, neuromorphic photonic hardware is energy-efficient but often lacks the ability to represent negative quantities natively. In this work, we propose a transformation method oriented to gated recurrent neural networks, such as Long Short Term Memory (LSTM) and Gated Recurrent Unit (GRU) models, converting them directly into deployable models, and targeting hardware accelerators that can only represent non-negative quantities. The proposed approach leverages pre-trained networks to efficiently generate equivalent non-negative models, eliminating the need for training non-negative models from scratch, a process that often leads to models with poor performance due to optimization challenges. The proposed method enables one to efficiently deploy gated RNNs to photonic hardware, while also offering additional understanding of model behavior, as highlighted in this work.","PeriodicalId":13135,"journal":{"name":"IEEE Transactions on Emerging Topics in Computational Intelligence","volume":"10 3","pages":"2471-2481"},"PeriodicalIF":6.0,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148514414","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Quantum Reachability Games","authors":"Wuniu Liu;Zhihui Li;Yongming Li","doi":"10.1109/TETCI.2024.3419704","DOIUrl":"https://doi.org/10.1109/TETCI.2024.3419704","url":null,"abstract":"The reachability of quantum games refers to the ability of two players to reach a certain subspace of Hilbert spaces in a quantum game modeled by a newly proposed quantum game process (QGP). We refer to this as quantum reachability games (QRGs). Reachability analysis is crucial in studying the strategic behavior and decision-making processes of players in quantum games, as it aids in devising optimal strategies to achieve desired objectives. In a QRG, players make decisions based on their quantum strategies related to decision epochs, which are represented by quantum operations or quantum measurements. We present an algorithm for evaluating two-player strategies, and investigate the optimal QRG for two players, where the goal of player 1 is to maximize his/her probability of reaching a given target subspace, while player 2 aims to minimize it. We show that any optimal QRG are determined, i.e., both of the two players have the same probability of winning in non-cooperative reachability games. Furthermore, our results also demonstrate that reachability games in the quantum world can be undecidable in certain cases. These findings are significantly distinct from those in the probabilistic or classical scenarios. Finally, we present three case studies, penny flipover, ambushing quantum walk and peer-to-peer scheme game on quantum network, to illustrate the potential applications and implications of QRGs.","PeriodicalId":13135,"journal":{"name":"IEEE Transactions on Emerging Topics in Computational Intelligence","volume":"10 3","pages":"2184-2198"},"PeriodicalIF":6.0,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148516073","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Privacy-Preserving Against Gradients Leakage Attacks via Joint Differential Privacy in Federated Learning","authors":"Yu Yang;Jianping Li;Xunyu Zhang","doi":"10.1109/TETCI.2026.3670862","DOIUrl":"https://doi.org/10.1109/TETCI.2026.3670862","url":null,"abstract":"Federated learning (FL) avoids direct data sharing by exchanging gradients, yet gradient leakage attacks can still reconstruct training samples and labels, particularly for neighbourhood-aware and continuously generated data. Existing differential privacy (DP) defences often incur substantial utility loss and do not simultaneously address record-level privacy for continually generated data and user-level privacy across participants. To bridge this gap, we propose Joint-DP-FL, a privacy-preserving FL framework that integrates joint differential privacy with three components: neighbourhood differential privacy, which groups semantically related records into neighbourhood domains and enables privacy-budget reuse for continuous data; gradient clipping to bound sensitivity and suppress abnormal updates; and Gaussian noise injection with privacy accounting to provide privacy guarantees at multiple granularities. Experiments on multiple models and image datasets show that Joint-DP-FL reduces the success rate of representative gradient-based leakage attacks while maintaining a favourable privacy-utility trade-off, and that neighbourhood-based budget reuse further lowers privacy consumption. Ablation and comparative studies illustrate the effect of each component on privacy protection and downstream task accuracy. Despite these gains, DP inevitably introduces performance degradation under strict privacy constraints, and the overall trade-off remains sensitive to noise calibration and clipping thresholds.","PeriodicalId":13135,"journal":{"name":"IEEE Transactions on Emerging Topics in Computational Intelligence","volume":"10 3","pages":"2456-2470"},"PeriodicalIF":6.0,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11456084","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148511242","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"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":"MGL-FBLS: Multi-Granularity Label-Driven Feature Enhanced Broad Learning System for Semi-Supervised Classification","authors":"Zhaoyin Shi;Yingjie Luo;Xin Liu;Long Chen;Weiping Ding;Xiaopin Zhong;Zongze Wu;C. L. Philip Chen","doi":"10.1109/TETCI.2025.3577454","DOIUrl":"https://doi.org/10.1109/TETCI.2025.3577454","url":null,"abstract":"The Broad Learning System (BLS) performs well in semi-supervised scenarios across many datasets, but its effectiveness is often limited by inadequate feature extraction and inefficient label utilization. To overcome these issues, we propose a Multi-Granularity Label-Driven Feature Enhanced Broad Learning System (MGL-FBLS) for Semi-Supervised Classification in this paper. First, for samples' feature enhancement, Random Fourier Features are imposed on the traditional broad model, which further strengthens the system's ability to capture nonlinear structures, without destroying the elegant and efficient randomness of BLS. It is worth noting that the model at this point can use exactly the same optimization method as the original model to handle fully supervised classification problems with higher accuracy efficiently. From the perspective of labels, our approach employs a multi-granularity labeling strategy that fully leverages both coarse and fine-granularity label information, enhancing the model's interpretability for unsupervised samples. In addition, a multi-manifold term is designed to describe the coarse-granularity semi-supervised information accurately and embed it into the sample feature-enhanced network efficiently. Meanwhile, a low-complexity, fast-converging algorithm is theoretically validated to maximize the use of both labeled and unlabeled data during training. Finally, comparative experiments in four benchmarks validate the proposed framework's effectiveness, demonstrating its competitive advantages over state-of-the-art methods.","PeriodicalId":13135,"journal":{"name":"IEEE Transactions on Emerging Topics in Computational Intelligence","volume":"10 3","pages":"2245-2259"},"PeriodicalIF":6.0,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148511346","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Taming Language Models for Predicting Stock Price With Causality Guidance","authors":"Yongcan Luo;Jiahao Zheng;Ning Chen;Dapeng Wu","doi":"10.1109/TETCI.2026.3663488","DOIUrl":"https://doi.org/10.1109/TETCI.2026.3663488","url":null,"abstract":"Predicting stock price trends not only needs qualitative analysis of textual information but also requires quantitative analysis of historical prices. Although recent advancements in language models have demonstrated their potential across various predictive tasks, current methods mainly focus on using a single modality with limited exploration of time series data in language models. To address both quantitative and qualitative analysis of stock price predicting tasks, we propose a multi-modality framework, the Joint Learning Time Series and Language Models (JLTSL) framework, which aligns time series information to language space. First, we encode time series data into temporal tokens, which can be used in joint training with textual tokens in language models. Second, we identify the presence of Granger Causality between time series sequences and their corresponding language representation. We utilize the Granger Causality test to select the prompt that has the strongest causality with the time series features to improve model performance. Our empirical evaluation across multiple datasets confirms that the JLTSL framework significantly outperforms existing benchmarks, underscoring its efficacy in predicting stock price trend.","PeriodicalId":13135,"journal":{"name":"IEEE Transactions on Emerging Topics in Computational Intelligence","volume":"10 3","pages":"2443-2455"},"PeriodicalIF":6.0,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148508769","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Self-Adaptive Differential Evolution-Enhanced Quality-Diversity Optimization for Distributed Flexible Job-Shop Scheduling With Transportation","authors":"Haoxiang Qin;Yi Xiang;Yuyan Han;Yuting Wang;Chunguo Wu;Fujian Feng","doi":"10.1109/TETCI.2026.3670689","DOIUrl":"https://doi.org/10.1109/TETCI.2026.3670689","url":null,"abstract":"Distributed Flexible Job Shop Scheduling with Transportation Constraints (DFJSP-T) is widely used in manufacturing workshops across various industries, as it more accurately represents real-world production scenarios. However, many existing studies do not fully leverage critical problem features or domain knowledge, limiting the ability of current algorithms to find high-quality solutions for complex large-scale problems. To address these challenges, this paper proposes a Self-Adaptive Differential Evolution Enhanced Quality-Diversity Optimization (SADE-QD). The SADE-QD incorporates domain knowledge in three main ways: (1) it models the problem using machine idle and job transportation features, allowing the algorithm to retain solutions with diverse behaviors; (2) it introduces a knowledge-guided heuristic search strategy based on critical path to discover more high-quality solutions; and (3) it applies an self-adaptive differential evolution search method that leverages machine idle and transportation features to explore a broader range of solutions. This helps valuable information from the feature space contribute directly to the search process. Experiments on small, medium, and large scale benchmarks show that SADE-QD achieves an average makespan reduction of 12% compared to several recent state-of-the-art algorithms.","PeriodicalId":13135,"journal":{"name":"IEEE Transactions on Emerging Topics in Computational Intelligence","volume":"10 3","pages":"2260-2275"},"PeriodicalIF":6.0,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148511411","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}