Pub Date : 2025-01-28DOI: 10.1016/j.ins.2025.121912
Artur Kasymov , Bartosz Czekaj , Marcin Mazur , Jacek Tabor , Przemysław Spurek
Gaussian Splatting involves embedding information about the desired 3D objects into multiple Gaussian distributions, which can be represented in 3D similarly to conventional meshes. It is regrettable that the use of Gaussians in Gaussian Splatting is currently somewhat restrictive due to their perceived simple nature. In practice, 3D objects are often composed of complex curves and highly twisted structures. This issue can to some extent be alleviated by employing a multitude of Gaussian components to accurately reflect the complex structures. However, this approach results in a considerable increase in time complexity. This paper introduces the concept of negative Gaussians, which are interpreted as items with negative colors. The rationale behind this approach is based on the density distribution created by subtracting the probability density functions (PDFs) of two Gaussians, which we refer to as Diff-Gaussian. This distribution can be used to approximate structures, such as donut-shaped and moon-shaped objects. Experimental findings indicate that the application of these techniques enhances the modeling of elements with high-frequency color transitions. Additionally, it improves the representation of shadows.
{"title":"NegGS: Negative Gaussian Splatting","authors":"Artur Kasymov , Bartosz Czekaj , Marcin Mazur , Jacek Tabor , Przemysław Spurek","doi":"10.1016/j.ins.2025.121912","DOIUrl":"10.1016/j.ins.2025.121912","url":null,"abstract":"<div><div>Gaussian Splatting involves embedding information about the desired 3D objects into multiple Gaussian distributions, which can be represented in 3D similarly to conventional meshes. It is regrettable that the use of Gaussians in Gaussian Splatting is currently somewhat restrictive due to their perceived simple nature. In practice, 3D objects are often composed of complex curves and highly twisted structures. This issue can to some extent be alleviated by employing a multitude of Gaussian components to accurately reflect the complex structures. However, this approach results in a considerable increase in time complexity. This paper introduces the concept of negative Gaussians, which are interpreted as items with negative colors. The rationale behind this approach is based on the density distribution created by subtracting the probability density functions (PDFs) of two Gaussians, which we refer to as Diff-Gaussian. This distribution can be used to approximate structures, such as donut-shaped and moon-shaped objects. Experimental findings indicate that the application of these techniques enhances the modeling of elements with high-frequency color transitions. Additionally, it improves the representation of shadows.</div></div>","PeriodicalId":51063,"journal":{"name":"Information Sciences","volume":"702 ","pages":"Article 121912"},"PeriodicalIF":8.1,"publicationDate":"2025-01-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143100042","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Legal case retrieval is a critical task in intelligent legal systems, providing relevant precedents to assist judges in their decision-making. While current data-driven neural retrieval methods have demonstrated impressive performance on clean, annotated data, they often ignore the robustness against noisy correspondences. In practice, legal annotators are required to identify legal uncertainty, which refers to the ambiguity or unpredictability in legal interpretations and applications, in relevance estimation between cases. This uncertainty often introduces noise into the training data, leading to unreliable predictions and potentially impacting the fairness and justice of downstream tasks. Focusing on this robustness issue, we propose a novel evidential learning framework called ELCR, which explicitly models the legal uncertainty and addresses noisy correspondences. Specifically, we first estimate the multi-faceted relevance between query-candidate cases from the concept, rule, and fact levels. These relevance estimations are then used to obtain the evidence-based uncertainty under the Dempster-Shafer Evidence Theory, which helps correct labels from noisy correspondence. Guided by two elaborate evidence-based training objectives, ELCR provides accurate uncertainty estimation, enhancing reliability and robustness. Extensive experiments on various noise proportions across two benchmark datasets demonstrate that our method exhibits robustness against noisy correspondences while maintaining competitive retrieval performance.
{"title":"Uncertainty-aware evidential learning for legal case retrieval with noisy correspondence","authors":"Weicong Qin , Weijie Yu , Kepu Zhang , Haiyuan Zhao , Jun Xu , Ji-Rong Wen","doi":"10.1016/j.ins.2025.121915","DOIUrl":"10.1016/j.ins.2025.121915","url":null,"abstract":"<div><div>Legal case retrieval is a critical task in intelligent legal systems, providing relevant precedents to assist judges in their decision-making. While current data-driven neural retrieval methods have demonstrated impressive performance on clean, annotated data, they often ignore the robustness against noisy correspondences. In practice, legal annotators are required to identify <em>legal uncertainty</em>, which refers to the ambiguity or unpredictability in legal interpretations and applications, in relevance estimation between cases. This uncertainty often introduces noise into the training data, leading to unreliable predictions and potentially impacting the fairness and justice of downstream tasks. Focusing on this robustness issue, we propose a novel evidential learning framework called ELCR, which explicitly models the <em>legal uncertainty</em> and addresses noisy correspondences. Specifically, we first estimate the multi-faceted relevance between query-candidate cases from the concept, rule, and fact levels. These relevance estimations are then used to obtain the evidence-based uncertainty under the Dempster-Shafer Evidence Theory, which helps correct labels from noisy correspondence. Guided by two elaborate evidence-based training objectives, ELCR provides accurate uncertainty estimation, enhancing reliability and robustness. Extensive experiments on various noise proportions across two benchmark datasets demonstrate that our method exhibits robustness against noisy correspondences while maintaining competitive retrieval performance.</div></div>","PeriodicalId":51063,"journal":{"name":"Information Sciences","volume":"702 ","pages":"Article 121915"},"PeriodicalIF":8.1,"publicationDate":"2025-01-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143100044","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-28DOI: 10.1016/j.ins.2025.121896
Lingna Shi , Jun-Guo Lu , Jiarong Li , Haijun Jiang , Jinling Wang , Yue Ren
This article explores the quasi-projective synchronization of multi-layer coupled neural networks employing pinning impulsive control. First, the network model incorporates intra- and inter-layer couplings while accounting for practical factors such as stochastic disturbances, leakage delay, and heterogeneous nodes. Second, to reduce control costs, we propose a pinning impulsive control strategy that applies impulses to key nodes. Moreover, a delayed pinning impulsive strategy is developed to address the potential delay in the controller's response. Then, utilizing stochastic differential equations and the comparison principle, quasi-projective synchronization conditions are gained and error bounds are accurately computed. Finally, numerical examples involving three-layer networks, along with comparative experiments, are provided to validate the theoretical findings.
{"title":"Pinning impulsive control for quasi-projective synchronization of stochastic multi-layer networks","authors":"Lingna Shi , Jun-Guo Lu , Jiarong Li , Haijun Jiang , Jinling Wang , Yue Ren","doi":"10.1016/j.ins.2025.121896","DOIUrl":"10.1016/j.ins.2025.121896","url":null,"abstract":"<div><div>This article explores the quasi-projective synchronization of multi-layer coupled neural networks employing pinning impulsive control. First, the network model incorporates intra- and inter-layer couplings while accounting for practical factors such as stochastic disturbances, leakage delay, and heterogeneous nodes. Second, to reduce control costs, we propose a pinning impulsive control strategy that applies impulses to key nodes. Moreover, a delayed pinning impulsive strategy is developed to address the potential delay in the controller's response. Then, utilizing stochastic differential equations and the comparison principle, quasi-projective synchronization conditions are gained and error bounds are accurately computed. Finally, numerical examples involving three-layer networks, along with comparative experiments, are provided to validate the theoretical findings.</div></div>","PeriodicalId":51063,"journal":{"name":"Information Sciences","volume":"702 ","pages":"Article 121896"},"PeriodicalIF":8.1,"publicationDate":"2025-01-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143100205","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-28DOI: 10.1016/j.ins.2025.121897
Mengyuan Li , Xiaohong Zhang , Haojie Jiang , Jun Liu
T-norms, t-conorms, uninorms, grouping functions, overlap functions, etc., are important fuzzy logic operators, they have been widely used in fuzzy reasoning, fuzzy control, information fusion, intelligent decision-making and fuzzy neural network. Recently, as a unified form of 1-grouping functions and 0-overlap functions, the new concept of function has been proposed. It is a new class of fuzzy logic operators with strong expressive power. However, we find that the parameter k in functions only belongs to {0,1} rather than [0,1], which limits their application scope. This article first delves into the characteristics of functions and provides several new construction theorems for functions. Then, more extensive OG-functions are proposed, proving that OG-functions are joint extension of the general grouping functions and general overlap functions. Multiple methods for constructing OG-functions are provided, and the structural theorem of OG-functions is proved (i.e., the necessary and sufficient conditions for generating OG-functions from “continuous symmetric nondecreasing function pairs”). Thirdly, OG-functions are extended to ()-OG functions, and a novel neuron model based on ()-OG functions (OG-neuron) is proposed for the first time. We also demonstrate OG-neurons have stronger approximation ability than traditional MP neurons (a single OG-neuron can achieve XOR operation). Finally, we establish novel artificial neural network OG-ANN and convolutional neural network OG-CNN. Comparative experimental results show that the introduction of ()-OG functions improves the classification accuracy of neural networks by 5.23%, 6.02%, 7.77% in mnist, cifar10 and fashion datasets, respectively.
{"title":"Some novel fuzzy logic operators with applications in fuzzy neural networks","authors":"Mengyuan Li , Xiaohong Zhang , Haojie Jiang , Jun Liu","doi":"10.1016/j.ins.2025.121897","DOIUrl":"10.1016/j.ins.2025.121897","url":null,"abstract":"<div><div>T-norms, t-conorms, uninorms, grouping functions, overlap functions, etc., are important fuzzy logic operators, they have been widely used in fuzzy reasoning, fuzzy control, information fusion, intelligent decision-making and fuzzy neural network. Recently, as a unified form of 1-grouping functions and 0-overlap functions, the new concept of <span><math><mi>Θ</mi><mo>−</mo><mi>Ξ</mi></math></span> function has been proposed. It is a new class of fuzzy logic operators with strong expressive power. However, we find that the parameter <em>k</em> in <span><math><mi>Θ</mi><mo>−</mo><mi>Ξ</mi></math></span> functions only belongs to {0,1} rather than [0,1], which limits their application scope. This article first delves into the characteristics of <span><math><mi>Θ</mi><mo>−</mo><mi>Ξ</mi></math></span> functions and provides several new construction theorems for <span><math><mi>Θ</mi><mo>−</mo><mi>Ξ</mi></math></span> functions. Then, more extensive <em>OG</em>-functions are proposed, proving that <em>OG</em>-functions are joint extension of the general grouping functions and general overlap functions. Multiple methods for constructing <em>OG</em>-functions are provided, and the structural theorem of <em>OG</em>-functions is proved (i.e., the necessary and sufficient conditions for generating <em>OG</em>-functions from “continuous symmetric nondecreasing function pairs”). Thirdly, <em>OG</em>-functions are extended to (<span><math><mi>a</mi><mo>,</mo><mi>b</mi></math></span>)-<em>OG</em> functions, and a novel neuron model based on (<span><math><mi>a</mi><mo>,</mo><mi>b</mi></math></span>)-<em>OG</em> functions (<em>OG</em>-neuron) is proposed for the first time. We also demonstrate <em>OG</em>-neurons have stronger approximation ability than traditional MP neurons (a single <em>OG</em>-neuron can achieve XOR operation). Finally, we establish novel artificial neural network <em>OG</em>-ANN and convolutional neural network <em>OG</em>-CNN. Comparative experimental results show that the introduction of (<span><math><mi>a</mi><mo>,</mo><mi>b</mi></math></span>)-<em>OG</em> functions improves the classification accuracy of neural networks by 5.23%, 6.02%, 7.77% in mnist, cifar10 and fashion datasets, respectively.</div></div>","PeriodicalId":51063,"journal":{"name":"Information Sciences","volume":"702 ","pages":"Article 121897"},"PeriodicalIF":8.1,"publicationDate":"2025-01-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143100045","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-27DOI: 10.1016/j.ins.2025.121902
Suhuan Zhang , Fanglai Zhu , Xufeng Ling
Deception attacks are regarded as the most typical cyber attacks, which may damage system performance or even cause system paralysis. Deception signals are well-designed by the attackers and may occur on any communication channel, which makes deception attacks relatively stealthy and difficult to detect. In this paper, for linear discrete-time systems that are vulnerable to deception attacks randomly occurring on the controller-to-actuator channel, a novel event-triggered unknown input observer (EUIO) is developed to suppress the impact of deception attacks and unknown inputs on the system and save communication resources. Firstly, different from existing results, the deception signal together with the system unknown input (UI) is modeled as a new multiple UI (MUI). Then, an algebraic expression of the MUI and the system state is established by designing an event-triggered interval observer (EIO). Subsequently, using the algebraic relationship, an EUIO is designed to asymptotically estimate the system state and the MUI simultaneously. Furthermore, an EUIO-based compensation controller is designed by incorporating both state estimation and MUI reconstruction (MUIR) to eliminate the influence of MUI on the system. Under the asymptotic convergence property of the estimations of the EUIO and the significant result of MUIR decoupling the control input, the stability of the closed-loop system can be ensured even if the system suffers from deception attacks, external disturbances, actuator faults, etc. Finally, the feasibility, effectiveness, and advantages of both the EUIO and EUIO-based controller are verified by two examples and comparisons.
{"title":"Event-triggered UIO-based security control for discrete-time systems under deception attacks","authors":"Suhuan Zhang , Fanglai Zhu , Xufeng Ling","doi":"10.1016/j.ins.2025.121902","DOIUrl":"10.1016/j.ins.2025.121902","url":null,"abstract":"<div><div>Deception attacks are regarded as the most typical cyber attacks, which may damage system performance or even cause system paralysis. Deception signals are well-designed by the attackers and may occur on any communication channel, which makes deception attacks relatively stealthy and difficult to detect. In this paper, for linear discrete-time systems that are vulnerable to deception attacks randomly occurring on the controller-to-actuator channel, a novel event-triggered unknown input observer (EUIO) is developed to suppress the impact of deception attacks and unknown inputs on the system and save communication resources. Firstly, different from existing results, the deception signal together with the system unknown input (UI) is modeled as a new multiple UI (MUI). Then, an algebraic expression of the MUI and the system state is established by designing an event-triggered interval observer (EIO). Subsequently, using the algebraic relationship, an EUIO is designed to asymptotically estimate the system state and the MUI simultaneously. Furthermore, an EUIO-based compensation controller is designed by incorporating both state estimation and MUI reconstruction (MUIR) to eliminate the influence of MUI on the system. Under the asymptotic convergence property of the estimations of the EUIO and the significant result of MUIR decoupling the control input, the stability of the closed-loop system can be ensured even if the system suffers from deception attacks, external disturbances, actuator faults, etc. Finally, the feasibility, effectiveness, and advantages of both the EUIO and EUIO-based controller are verified by two examples and comparisons.</div></div>","PeriodicalId":51063,"journal":{"name":"Information Sciences","volume":"702 ","pages":"Article 121902"},"PeriodicalIF":8.1,"publicationDate":"2025-01-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143100208","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-27DOI: 10.1016/j.ins.2025.121906
Limin Hua, Han Liu, Yinghui Pan
A Job Shop Scheduling Problem (JSSP) is an NP-Hard problem with extensive applications in many domains such as transportation and manufacturing industrial chains. Deep Reinforcement Learning (DRL) has emerged as a novel approach being distinct from traditional scheduling and heuristic methods. Although DRL has shown promising results in addressing JSSP, several limitations remain, such as ignoring an optimal solution space and lacking the focus in policy network learning, which affects both scheduling quality and learning speed. To address these issues, we introduce the DecayP30 method that incorporates the solution space partition and dynamic weight allocation into the decision-making process. Specifically, the DecayP30 method innovatively replaces the traditional clipping operation in Proximal Policy Optimization (PPO) with the Sigmoid function, a key feature of the Preconditioner Proximal Policy Optimization (P30) approach. We introduce a dynamic decay strategy to address the “heavy-head and light-tail” issue JSSP. The new approach ensures a more comprehensive solution space while emphasizing sequential relationships inherent in JSSP. We evaluate the new method in four major JSSP datasets. Extensive experiments demonstrate that our proposed method exhibits better convergence speed and scheduling quality compared to most the DRL methods.
{"title":"Solving industrial chain job scheduling problems through a deep reinforcement learning method with decay strategy","authors":"Limin Hua, Han Liu, Yinghui Pan","doi":"10.1016/j.ins.2025.121906","DOIUrl":"10.1016/j.ins.2025.121906","url":null,"abstract":"<div><div>A Job Shop Scheduling Problem (JSSP) is an NP-Hard problem with extensive applications in many domains such as transportation and manufacturing industrial chains. Deep Reinforcement Learning (DRL) has emerged as a novel approach being distinct from traditional scheduling and heuristic methods. Although DRL has shown promising results in addressing JSSP, several limitations remain, such as ignoring an optimal solution space and lacking the focus in policy network learning, which affects both scheduling quality and learning speed. To address these issues, we introduce the DecayP30 method that incorporates the solution space partition and dynamic weight allocation into the decision-making process. Specifically, the DecayP30 method innovatively replaces the traditional clipping operation in Proximal Policy Optimization (PPO) with the Sigmoid function, a key feature of the Preconditioner Proximal Policy Optimization (P30) approach. We introduce a dynamic decay strategy to address the “heavy-head and light-tail” issue JSSP. The new approach ensures a more comprehensive solution space while emphasizing sequential relationships inherent in JSSP. We evaluate the new method in four major JSSP datasets. Extensive experiments demonstrate that our proposed method exhibits better convergence speed and scheduling quality compared to most the DRL methods.</div></div>","PeriodicalId":51063,"journal":{"name":"Information Sciences","volume":"702 ","pages":"Article 121906"},"PeriodicalIF":8.1,"publicationDate":"2025-01-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143100204","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Face anti-spoofing (FAS) aims to detect and counter facial presentation attacks by distinguishing subtle differences between live and spoof faces. Since the vision transformer (ViT) has demonstrated significant performance gains in various computer vision tasks by effectively capturing long-range dependencies, recent FAS methods have investigated the potential of leveraging the channel-wise features derived from self-attention (SA) in ViT. While channel-wise features effectively capture discriminative local attentions, the complementary information existing between different channels remained undiscovered in prior methods. In this paper, we investigate the unexplored characteristics of complementary channel information within ViT and propose to incorporate both channel-wise and complementary channel information to learn long-range and discriminative features for FAS. We design two modules, including Channel Difference Self-Attention (CDSA) and Multi-head Channel Difference Self-Attention (MCDSA), to facilitate learning complementary channel characteristics and enhancing both feature discriminability and representational capacity. Building upon CDSA and MCDSA, we propose a novel and efficient Channel Difference Transformer (CDformer) without introducing any additional parameters or increasing computation complexity of ViT. Extensive experiments conducted on five FAS benchmark datasets demonstrate that our proposed CDformer achieves state-of-the-art performance on both intra-domain and cross-domain testing scenarios.
{"title":"Channel difference transformer for face anti-spoofing","authors":"Pei-Kai Huang , Jun-Xiong Chong , Ming-Tsung Hsu , Fang-Yu Hsu , Chiou-Ting Hsu","doi":"10.1016/j.ins.2025.121904","DOIUrl":"10.1016/j.ins.2025.121904","url":null,"abstract":"<div><div>Face anti-spoofing (FAS) aims to detect and counter facial presentation attacks by distinguishing subtle differences between live and spoof faces. Since the vision transformer (ViT) has demonstrated significant performance gains in various computer vision tasks by effectively capturing long-range dependencies, recent FAS methods have investigated the potential of leveraging the channel-wise features derived from self-attention (SA) in ViT. While channel-wise features effectively capture discriminative local attentions, the complementary information existing between different channels remained undiscovered in prior methods. In this paper, we investigate the unexplored characteristics of complementary channel information within ViT and propose to incorporate both channel-wise and complementary channel information to learn long-range and discriminative features for FAS. We design two modules, including Channel Difference Self-Attention (CDSA) and Multi-head Channel Difference Self-Attention (MCDSA), to facilitate learning complementary channel characteristics and enhancing both feature discriminability and representational capacity. Building upon CDSA and MCDSA, we propose a novel and efficient Channel Difference Transformer (CDformer) without introducing any additional parameters or increasing computation complexity of ViT. Extensive experiments conducted on five FAS benchmark datasets demonstrate that our proposed CDformer achieves state-of-the-art performance on both intra-domain and cross-domain testing scenarios.</div></div>","PeriodicalId":51063,"journal":{"name":"Information Sciences","volume":"702 ","pages":"Article 121904"},"PeriodicalIF":8.1,"publicationDate":"2025-01-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143100041","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Public Key Encryption with Keyword Search (PEKS) is a promising technique that enables secure search over encrypted data. However, recent legislation mandates specific activation times for information, necessitating time-controlled retrieval. This creates a pressing need to integrate timed-release control into PEKS, allowing data uploaders to “send indices to the future”, ensuring that secure, searchable indices for specific keywords can only be searched after the designated release-time. Existing approaches or straightforward candidates have several limitations: they either lack cryptographic search control and precise policy enforcement, fail to support flexible and efficient policy adjustment, or exhibit inefficiencies in index/key size and search complexity. In this paper, we formalize a novel variant called Fully-Incremental Public Key Encryption with Timed-Release Keyword Search (Fi-PETRKS), which well captures the functionalities, efficiency, and security requirements. Notably, the processes of functionalities are all incremental, ensuring that the size of token used for each adjustment remains . We propose a concrete Fi-PETRKS construction which is secure against full keyword guessing attacks. Furthermore, we introduce an enhanced version, Fi-PETRKS+, which offers sub-linear search efficiency. Both theoretical analysis and experimental results demonstrate the practicality of our scheme.
{"title":"Fully-incremental public key encryption with adjustable timed-release keyword search","authors":"Tiancheng Zhu , Jiabei Wang , Yuting Xiao , Yiwen Gao , Yongbin Zhou , Jian Weng","doi":"10.1016/j.ins.2025.121887","DOIUrl":"10.1016/j.ins.2025.121887","url":null,"abstract":"<div><div>Public Key Encryption with Keyword Search (PEKS) is a promising technique that enables secure search over encrypted data. However, recent legislation mandates specific activation times for information, necessitating time-controlled retrieval. This creates a pressing need to integrate timed-release control into PEKS, allowing data uploaders to “<em>send indices to the future</em>”, ensuring that secure, searchable indices for specific keywords can only be searched after the designated release-time. Existing approaches or straightforward candidates have several limitations: they either lack cryptographic search control and precise policy enforcement, fail to support flexible and efficient policy adjustment, or exhibit inefficiencies in index/key size and search complexity. In this paper, we formalize a novel variant called Fully-Incremental Public Key Encryption with Timed-Release Keyword Search (<span>Fi-PETRKS</span>), which well captures the functionalities, efficiency, and security requirements. Notably, the processes of functionalities are all incremental, ensuring that the size of token used for each adjustment remains <span><math><mi>O</mi><mo>(</mo><mn>1</mn><mo>)</mo></math></span>. We propose a concrete <span>Fi-PETRKS</span> construction which is secure against full keyword guessing attacks. Furthermore, we introduce an enhanced version, <span>Fi-PETRKS+</span>, which offers sub-linear search efficiency. Both theoretical analysis and experimental results demonstrate the practicality of our scheme.</div></div>","PeriodicalId":51063,"journal":{"name":"Information Sciences","volume":"702 ","pages":"Article 121887"},"PeriodicalIF":8.1,"publicationDate":"2025-01-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143156641","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-27DOI: 10.1016/j.ins.2025.121903
Shun Zheng, Shaoqing Wang, Keke Li, Xueting Li, Fuzhen Sun
The key to sequential recommendations is to learn accurate item embeddings based on users' historical interactions. Existing methods rely on either fixed vectors or distributions to represent items. Though these methods are effective, we argue there are two limitations. a) User preferences for items can arise from multiple aspects. Fixed vectors are insufficient to capture the multifaceted features of items and the user's novel intentions. b) The conventional constraints of distribution representations during modeling process lead to the loss of some flexibility. To address these limitations, we propose the Hybrid Representation model for Sequential Recommendation (HR4SR), which utilizes both fixed vectors and distribution representations to model the features in interaction sequences. Specifically, we propose the use of diffusion techniques to introduce distribution vectors by gradually adding noise, which can compensate for the inadequacies of fixed vectors. In addition, we remove the traditional constraints of diffusion techniques, allowing distribution vectors to capture the fine-grained features of items more flexibly. Finally, we utilize a combination of fixed vectors and distributed vectors to form the item embeddings, where distributed vectors are means to compensate for the fixed vectors' inability to capture finer-grained user preferences. Experiments on three datasets show that HR4SR significantly outperforms strong baselines. The code is released at https://github.com/xiaocilu1999/HR4SR.
{"title":"When feature encoder meets diffusion model for sequential recommendations","authors":"Shun Zheng, Shaoqing Wang, Keke Li, Xueting Li, Fuzhen Sun","doi":"10.1016/j.ins.2025.121903","DOIUrl":"10.1016/j.ins.2025.121903","url":null,"abstract":"<div><div>The key to sequential recommendations is to learn accurate item embeddings based on users' historical interactions. Existing methods rely on either fixed vectors or distributions to represent items. Though these methods are effective, we argue there are two limitations. a) User preferences for items can arise from multiple aspects. Fixed vectors are insufficient to capture the multifaceted features of items and the user's novel intentions. b) The conventional constraints of distribution representations during modeling process lead to the loss of some flexibility. To address these limitations, we propose the Hybrid Representation model for Sequential Recommendation (HR4SR), which utilizes both fixed vectors and distribution representations to model the features in interaction sequences. Specifically, we propose the use of diffusion techniques to introduce distribution vectors by gradually adding noise, which can compensate for the inadequacies of fixed vectors. In addition, we remove the traditional constraints of diffusion techniques, allowing distribution vectors to capture the fine-grained features of items more flexibly. Finally, we utilize a combination of fixed vectors and distributed vectors to form the item embeddings, where distributed vectors are means to compensate for the fixed vectors' inability to capture finer-grained user preferences. Experiments on three datasets show that HR4SR significantly outperforms strong baselines. The code is released at <span><span>https://github.com/xiaocilu1999/HR4SR</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":51063,"journal":{"name":"Information Sciences","volume":"702 ","pages":"Article 121903"},"PeriodicalIF":8.1,"publicationDate":"2025-01-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143100047","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-27DOI: 10.1016/j.ins.2025.121886
Shu Liu , Huaguang Zhang , Jiayue Sun
This paper addresses the consensus problem of linear multi-agent systems (MASs) in directed graphs using a timer-based event-triggered control algorithm. The proposed distributed algorithm allows each agent to update its control law and event monitoring condition using only relative state information from neighboring agents at discrete event instants. This algorithm minimizes reliance on global information and significantly reduces communication overhead, thereby enhancing both efficiency and scalability. A common challenge in event-based control algorithms is the potential for Zeno behavior, where an infinite number of events could occur within a finite time, making the system impractical. While conventional algorithms avoid Zeno behavior by ensuring non-zero time intervals between events, they often fail to address the issue of excessively short event intervals. Our algorithm overcomes this limitation by establishing a strictly positive lower bound for the interval between events for each agent, thereby not only avoiding Zeno behavior but also ensuring practical applicability and robustness of the control strategy. Through simulation studies, we validate the efficacy of our algorithm in achieving asymptotic consensus in linear MASs over directed graphs.
{"title":"Timer-based distributed coordination for achieving asymptotic consensus in directed communication networks","authors":"Shu Liu , Huaguang Zhang , Jiayue Sun","doi":"10.1016/j.ins.2025.121886","DOIUrl":"10.1016/j.ins.2025.121886","url":null,"abstract":"<div><div>This paper addresses the consensus problem of linear multi-agent systems (MASs) in directed graphs using a timer-based event-triggered control algorithm. The proposed distributed algorithm allows each agent to update its control law and event monitoring condition using only relative state information from neighboring agents at discrete event instants. This algorithm minimizes reliance on global information and significantly reduces communication overhead, thereby enhancing both efficiency and scalability. A common challenge in event-based control algorithms is the potential for Zeno behavior, where an infinite number of events could occur within a finite time, making the system impractical. While conventional algorithms avoid Zeno behavior by ensuring non-zero time intervals between events, they often fail to address the issue of excessively short event intervals. Our algorithm overcomes this limitation by establishing a strictly positive lower bound for the interval between events for each agent, thereby not only avoiding Zeno behavior but also ensuring practical applicability and robustness of the control strategy. Through simulation studies, we validate the efficacy of our algorithm in achieving asymptotic consensus in linear MASs over directed graphs.</div></div>","PeriodicalId":51063,"journal":{"name":"Information Sciences","volume":"703 ","pages":"Article 121886"},"PeriodicalIF":8.1,"publicationDate":"2025-01-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143103850","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号