Information Sciences最新文献

In this article, an adaptive optimized consensus tracking control problem is studied for nonlinear strict-feedback dynamic multi-agent systems (MASs), considering both unmeasurable system states and time-varying bias faults. By utilizing the backstepping technique, we develop an adaptive reinforcement learning (RL) algorithm within the observer-critic-actor architecture, specially designed to compensate for the lack of state information and derive control inputs, thereby achieving approximate optimal control. Moreover, an event-triggered mechanism is introduced in the sensor-to-controller channel, which dynamically adjusts the triggering threshold online and employs event-sampled states to initiate control actions. To address discontinuities caused by state triggering, we construct virtual controllers that continuously sample state signals and reconfigure the actual controller based on previously triggered states. The outputs of the MASs are shown to accurately track the desired reference signals while ensuring the boundedness of all closed-loop signals. Additionally, the proposed controller is verified to be devoid of Zeno behavior. Finally, the effectiveness of our control methodology is demonstrated through numerical simulation.

In predictive modeling, quantifying prediction uncertainty is crucial for reliable decision-making. Traditional conformal inference methods provide marginally valid predictive regions but often produce non-adaptive intervals when naively applied to regression, potentially biasing applications. Recent advances using quantile regressors or conditional density estimators improve adaptability but are typically tied to specific prediction models, limiting their ability to quantify uncertainty around arbitrary models. Similarly, methods based on partitioning the feature space adopt sub-optimal strategies, failing to consistently measure predictive uncertainty across the feature space, especially in adversarial examples. This paper introduces a model-agnostic family of methods to calibrate prediction intervals for regression with local coverage guarantees. By leveraging regression trees and Random Forests, our approach constructs data-adaptive partitions of the feature space to approximate conditional coverage, enhancing the accuracy and scalability of prediction intervals. Our methods outperform established benchmarks on simulated and real-world datasets. They are implemented in the Python package clover, which integrates seamlessly with the scikit-learn interface for practical application.

Interpretability is the next frontier in machine learning research. In the search for white box models — as opposed to black box models, like random forests or neural networks — rule induction algorithms are a logical and promising option, since the rules can easily be understood by humans. Fuzzy and rough set theory have been successfully applied to this archetype, almost always separately. As both approaches offer different ways to deal with imprecise and uncertain information, often with the use of an indiscernibility relation, it is natural to combine them. The QuickRules [20] algorithm was a first attempt at using fuzzy rough set theory for rule induction. It is based on QuickReduct, a greedy algorithm for building decision superreducts. QuickRules already showed an improvement over other rule induction methods. However, to evaluate the full potential of a fuzzy rough rule induction algorithm, one needs to start from the foundations. Accordingly, the novel rule induction algorithm, Fuzzy Rough Rule Induction (FRRI), we introduce in this paper, uses an approach that has not yet been utilised in this setting. We provide background and explain the workings of our algorithm. Furthermore, we perform a computational experiment to evaluate the performance of our algorithm and compare it to other state-of-the-art rule induction approaches. We find that our algorithm is more accurate while creating small rulesets consisting of relatively short rules.

The substantial proliferation of software vulnerabilities poses a persistent threat to system security, driving increased interest in applying deep learning (DL) for vulnerability detection. However, DL-based detectors often operate with a fixed number of input tokens, leading to semantic loss over large code snippets. Additionally, developing these detectors demands substantial labeled data and training time. To address these limitations, this paper proposes HardVD, which explores High-capacity cross-modal adversarial reprogramming for data-efficient Vulnerability Detection. HardVD devises a high-capacity semantic extractor to capture salient features per line of code, which are then arranged as patches to form an image representing the target function. These images are processed using convolutional filters as universal perturbations and non-parametric label remapping to adapt a pretrained Vision Transformer (ViT) for vulnerability detection, updating only the limited parameters of the perturbation filters during training. Extensive experiments demonstrate that HardVD outperforms DL-based baselines in terms of detection effectiveness, data-limited performance, and computational overhead. The ablation study also confirms the essential role of our high-capacity semantic extractor, without which an averaged relative decrease of 5.87% and 7.98% in accuracy and F1 score is observed, respectively.

Since large-scale multi-objective problems (LSMOPs) have huge decision variables, the traditional evolutionary algorithms are facing difficulties of low exploitation efficiency and high exploration costs in solving LSMOPs. Therefore, this paper proposes an evolutionary strategy based on two-stage accelerated search optimizers (ATAES). Specifically, a convergence optimizer is devised in the first stage, while a three-layer lightweight convolutional neural network model is built, and the population is homogenized into two subsets, the diversity subset, and the convergence subset, which serve as input nodes and the expected output nodes of the neural network, respectively. Then, by constantly backpropagating the gradient, a satisfactory individual will be produced. Once exploitation stagnation is discovered in the first phase, the second phase will be run, where a diversity optimizer using a differential optimization algorithm with opposite learning is suggested to increase the exploration range of candidate solutions and thereby increase the population's diversity. Finally, to validate the algorithm's performance, on multi-objective LSMOP and DTLZ benchmark suits with decision variable quantities of 100, 300, 500, and 1000, the ATAES demonstrated its superiority with other advanced multi-objective evolutionary algorithms.

Continual learning refers to the capability of a machine learning model to learn and adapt to new information, without compromising its performance on previously learned tasks. Although several studies have investigated continual learning methods for neural information retrieval (NIR) tasks, a well-defined task definition is still lacking, and it is unclear how typical learning strategies perform in this context. To address this challenge, a systematic task definition of continual NIR is presented, along with a multiple-topic dataset that simulates continuous information retrieval. A comprehensive continual neural information retrieval framework consisting of typical retrieval models and continual learning strategies is then proposed. Empirical evaluations illustrate that the proposed framework can successfully prevent catastrophic forgetting in neural information retrieval and enhance performance on previously learned tasks. The results also indicate that embedding-based retrieval models experience a decline in their continual learning performance as the topic shift distance and dataset volume of new tasks increase. In contrast, pretraining-based models do not show any such correlation. Adopting suitable learning strategies can mitigate the effects of topic shift and data augmentation in continual neural information retrieval.

In recent years, metric-based meta-learning methods have received widespread attention because of their effectiveness in solving few-shot classification problems. However, the scarcity of data frequently results in suboptimal embeddings, causing a discrepancy between anticipated class prototypes and those derived from the support set. These problems severely limit the generalizability of such methods, necessitating further development of Few-Shot Learning (FSL). In this study, we propose the Contrastive Prototype Network (CPN) consisting of three components: (1) Contrastive learning proposed as an auxiliary path to reduce the distance between homogeneous samples and amplify the differences between heterogeneous samples, thereby enhancing the effectiveness and quality of embeddings; (2) A pseudo-prototype strategy proposed to address the bias in prototypes, whereby the pseudo prototypes generated using query set samples are integrated with the initial prototypes to obtain more representative prototypes; (3) A new data augmentation technique, mixupPatch, introduced to alleviate the issue of insufficient data samples, whereby enhanced images are generated by blending the images and labels from different samples, to increase the number of samples. Extensive experiments and ablation studies conducted on five datasets demonstrated that CPN achieves robust results against recent solutions.

Highly detailed and accurate segmentation and classification of images constitutes an important class of tasks in computer vision. Typical “universal” domain-agnostic methods are known to suffer from instabilities and are prone to adversarial perturbations. Natural heterogeneity inherent in biological tissue structures complicates the interpretation of images even by trained physicians. Yet, algorithms in the medical domain require a high level of stability and interpretability to ensure their adoption by clinical experts and acceptance in clinical decision-making. In this work, we propose a novel method for segmentation and classification to address these challenges. The method is based on a hierarchical approach and biologically-informed feature extraction. The method's technical pipeline includes the automatic extraction of key biologically-informed features typically considered by physicians. This is followed by image classification using these features. Both stages rely on randomized ML techniques. The proposed hierarchical biomedically-informed approach significantly improved the image classification quality compared to the baseline solution of image classification in the task of colorectal cancer (CRC) analysis. The average F1-score for the four tissue types increased from 0.737 to 0.956. Using tumor tissue classification task as an example, we showed that the proposed algorithm offers an effective and practical avenue to solve these challenging issues.

Both classical forecasting methods and machine learning approaches are used to solve forecasting problems. Deep artificial neural networks, one of the machine learning methods, are widely used today and give very good results. Recurrent neural networks, a type of deep neural network, are very important in forecasting problems. Simple recurrent artificial neural networks, which are the simplest deep recurrent neural networks, are often preferred in solving forecasting problems due to the small number of parameters they use. Simple exponential smoothing, one of the classical forecasting methods, attracts attention with its performance in solving forecasting problems. The motivation of the study is to create a new forecasting method by combining a classical and simple forecasting method with a deep recurrent artificial neural network in an architecture. In this, a new hybrid deep recurrent artificial neural network with a simple exponential smoothing feedback mechanism is proposed. The architecture of the proposed method is created as a combination of simple recurrent artificial neural networks and simple exponential smoothing methods. In the training of the proposed method, two training algorithms based on sine cosine optimization and particle swarm optimization algorithms are proposed. In these training algorithms, two different solution strategies such as restarting, and early stopping rule are used to avoid overfitting and local optimum problems. The performance of the proposed method is analysed using stock market datasets and compared with both different deep and shallow artificial neural networks and classical forecasting methods. As a result of the analyses, it is concluded that the proposed method is successful in one step ahead of forecasting performance.

Verifiable data streaming (VDS) allows users to continuously upload their encrypted data items to untrusted cloud servers to reduce the burden of local storage. Meanwhile, it enables users to update outsourced data and initiate queries to verify data integrity. However, most previous works focus on queries for a certain index and are not compatible with general keyword queries. To this end, we propose a VDS protocol that supports keyword queries. Specifically, we first present an efficient dynamic symmetric searchable encryption (DSSE) with range query, which can represent multiple queried keywords within a range as the concise query token, thereby achieving communication-optimized keyword queries while achieving unbounded data appending. Furthermore, we construct a new VDS protocol supporting keyword queries by integrating the proposed range DSSE. Specifically, suppose the user searches for multiple keywords within a certain range. In that case, cloud servers can match the corresponding data items based on the user's query token, and users can verify the integrity of these data items. Security analysis demonstrates that our VDS protocol achieves forward security. Experimental results show that it sacrifices acceptable costs in exchange for keyword retrieval capability.