摘要:
Scan testing is widely used in the rigorous testing of modern integrated circuits. However, the controllable and observable nature of the scan design also provides opportunities for attackers who can utilise this structure to steal secret information stored in the chip. This paper proposes a secure Design for Testability (DFT) architecture to prevent scan-based attacks. The scheme uses a modified Linear Feedback Shift Register (LFSR) to dynamically generate keys for the scan design key generator and verifies them against a specific selection of test key flip-flops in the original scan chain, and locks the control signals by transmitting the outputs of the verification comparisons to a OR gate. When the test authorisation key is correct, the circuit performs a normal scan operation; otherwise, the circuit enters an abnormal test mode and the data in the scan chain is dynamically obfuscated, thus preventing from observing valid data or deriving an encryption key through the output of the scan chain.
摘要:
In this paper, inspired by a 3D multi-wing chaotic system proposed by Sahoo et al. in 2022, we presented a 5D multi-wing fractional-order memristive chaotic system with hidden attractors by increasing dimensionality and introducing memristor. Firstly, the equilibrium point of the 5D fractional-order memristive chaotic system was analyzed, and it was found that it has hidden attractors. The adomian decomposition algorithm was used to decompose the nonlinear term of the system. Then, the dynamic behavior of the system was analyzed using phase diagrams, Lyapunov exponent spectrums, and bifurcation diagrams. The hardware circuit of the 5D multi-wing fractional-order chaotic system was implemented using a field programmable gate array (FPGA) and experimentally verified. The experimental results are consistent with the numerical simulation results in MATLAB. Finally, the good randomness of the system was verified through NIST testing.
通讯机构:
[Zhang, JM ] C;Changsha Univ Sci & Technol, Sch Comp & Commun Engn, Changsha 410114, Peoples R China.
关键词:
Object tracking;Cross-correlation;Transformer;Decoupled head
摘要:
The fusion of the template and search region features plays a significant role in deep learningbased trackers. In Siamese -based trackers, different cross -correlation operations are commonly used to fuse features, which cannot obtain global connections. On the other hand, transformerbased trackers use attention mechanism to fuse features, which cannot suppress the interference of distractors in the background. Furthermore, existing trackers use regression and classification heads with the same structure, which leads to lack a deeper understanding of these two different tasks. To address these problems, we firstly propose a feature enhancement -fusion network (FEFN) based on cross -correlation and transformer, with two Encoders that employ self -attention and a Decoder that removes cross -attention to adapt to the tracking task. Using the FEFN to combine the advantages of Siamese -based and transformer -based trackers, our tracker establishes global connections while effectively suppressing the distractors. We also propose a novel decoupled head, designing a spatial sensitive classification head and a global information sensitive regression head, which helps the context -aware tracker locate the target more accurately. Our proposed tracker obtains 0.710 of AO, 0.814 of SR0.5 and 0.657 of SR0.75 on the GOT -10k test set, and achieves real-time requirement at 36.99FPS.
摘要:
For Deepfake detection, many existing works use the cross-entropy loss to enforce the classifier network to learn the mapping relationship from the RGB domain to the class domain, lacking an explicit constraint to guide the feature extraction network to learn discriminative features from an input image. This constrains the feature representation capability to expose deepfake. In this work, we analyze the feature extraction network in terms of both difference and similarity capabilities and propose a new constraint called similarity loss (SL) to improve the detection performance of the convolutional neural network (CNN) based detector. Moreover, according to the experimental results of the SL on data augmentation effectiveness, we propose a simple yet efficient framework, which is called as efficient similarity representation learning (ESRL), for deepfake detection. Extensive experiments on three public datasets (namely FF++, DFDC, and Celeb-DF) show that the feature extraction network trained with the help of SL can map forged faces and real faces to different feature embedding and map the same type of forged faces to similar feature embedding.
期刊:
Biomedical Signal Processing and Control,2024年92:106058 ISSN:1746-8094
通讯作者:
Jianming Zhang
作者机构:
[Li-Dan Kuang; Hao-Peng Zhang; Hao Zhu; Shiming He; Wenjun Li; Yan Gui; Jin Zhang; Jianming Zhang] School of Computer and Communication Engineering, Changsha University of Science and Technology, Changsha, 410114, China
通讯机构:
[Jianming Zhang] S;School of Computer and Communication Engineering, Changsha University of Science and Technology, Changsha, 410114, China
摘要:
The rank-(L, L, 1, 1) block term decomposition (BTD) model shows better separation performance for multi-subject fMRI data by preserving the high-way structure of fMRI data than canonical polyadic decomposition (CPD). However, multi-subject fMRI data are noisy and have high spatiotemporal variability. To address these limitations, this paper proposes a novel 3D weighted spatial pooling preprocessing that compresses and smooths multi-subject fMRI data and assigns a higher weight to in-brain voxels but a lower weight to out-brain voxels. This strategy not only largely reduces the size of spatial images but also improves the robustness to noise. Furthermore, to address the high spatiotemporal variability, the rank-(L, L, 1, 1) BTD model of the reduced fMRI data is relaxed by incorporating temporal shift-invariance and spatial orthonormality constraints to extract pooled multi-subject shared spatial maps, shared time courses, subject-specific time delays and intensities. Finally, multi-subject intact shared spatial maps are obtained based on shift-invariant rank-(L, L, 1, 1) BTD of intact fMRI data. The simulated and experimental fMRI data experiments both verify that the proposed method achieves better separation performance and stronger robustness to noise than rank-(L, L, 1, 1) BTD with a spatial orthonormality constraint and a method combining independent component analysis and shift-invariant CPD. Moreover, the proposed method with 3D spatial pooling yields better separation performance than that with 2D spatial pooling, because 3D spatial pooling preserves refined voxels, thereby retaining more information of adjacent slices.
摘要:
Images captured in environments with poor lighting conditions often suffer from insufficient brightness, significant noise, and color distortion, which is highly detrimental to subsequent high-level vision tasks. Low-light image enhancement requires effective feature extraction and fusion, and the advantages of transformer and convolution in image processing are complementary. Therefore, it is an intentional exploration to combine them in image enhancement. In this paper, we propose a novel UNet-like method for enhancing low-light images. Transformer blocks are stacked to form the encoder, and convolutional blocks are utilized in the decoder. First, considering that the Transformer can effectively capture global information and convolution can obtain local information, this paper improves the lightweight Transformer by integrating multi-scale depth-wise convolution into the feedforward network to extract comprehensive features. Then, we design a Skip Cross-Attention Module to replace the traditional skip connection, which combines feature maps from different stages of the encoder and decoder. To achieve better feature fusion, this module employs two masks: the Top-k Mask and the adaptive V Channel Mask based on maximum entropy. The Top-k Mask will filter out unfavorable features by preserving the top k scores of attention map, and the V Channel Mask utilizes the corrected V channel of the image as an illumination guide for enhancement. Finally, extensive experiments on seven datasets demonstrate that our method achieves good performance in both subjective and objective evaluations. Specifically, the runtime on LOL-v2-real dataset demonstrates that our method is close to achieving real-time performance.
期刊:
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS,2024年21:1-1 ISSN:1545-598X
作者机构:
[Honglin Wu; Peng Huang; Min Zhang; Wenlong Tang] School of Computer and Communication Engineering, Changsha University of Science and Technology, Changsha, China
摘要:
Remote-sensing image semantic segmentation is usually based on convolutional neural networks (CNNs). CNNs demonstrate powerful local feature extraction capabilities through stacked convolution and pooling. However, the locality of the convolution operation limits the ability of CNNs to directly extract global information. Relying on the multihead self-attention (MHSA) mechanism, transformer shows great advantages in modeling global information. In this letter, we propose a CNN-transformer fusion network (CTFNet) for remote-sensing image semantic segmentation. CTFNet applies a U-shaped encoder-decoder structure to achieve the extraction and adaptive fusion of local features and global context information. Specifically, a lightweight W/P transformer block is proposed as the decoder to obtain global context information with low complexity and connected to the encoder through the skip connection. Finally, the channel and spatial attention fusion module (AFM) is exploited to adaptively fuse deep semantic features and shallow detail features. On the Vaihingen and Potsdam datasets of the International Society for Photogrammetry and Remote Sensing (ISPRS), the effectiveness of each module is demonstrated by ablation experiments. Compared with several classical networks, our proposed CTFNet can obtain superior performance.
关键词:
radar reflectivity images;hail recognition;strong-echo areas;spatio-temporal radar image sequences;meteorological data
摘要:
Hail, an intense convective catastrophic weather, is seriously hazardous to people's lives and properties. This article proposes a multi-step cyclone hail weather recognition model, called long short-term memory (LSTM)-C3D, based on radar images, integrating attention mechanism and network voting optimization characteristics to achieve intelligent recognition and accurate classification of hailstorm weather based on long short-term memory networks. Based on radar echo data in the strong-echo region, LSTM-C3D can selectively fuse the long short-term time feature information of hail meteorological images and effectively focus on the significant features to achieve intelligent recognition of hail disaster weather. The meteorological scans of 11 Doppler weather radars deployed in various regions of the Hunan Province of China are used as the specific experimental and application objects for extensive validation and comparison experiments. The results show that the proposed method can realize the automatic extraction of radar reflectivity image features, and the accuracy of hail identification in the strong-echo region reaches 91.3%. It can also effectively realize the prediction of convective storm movement trends, laying the theoretical foundation for reducing the misjudgment of extreme disaster weather.
摘要:
Building artificial neural network models and studying their dynamic behaviors is extremely important from both a theoretical and practical standpoint due to the rapid advancement of artificial intelligence . In addition to its engineering applications, this article concentrates primarily on the memristor model and chaotic dynamics of the asymmetric memristive neural network. First, we develop a novel memristor model, which is multistable and highly tunable. Using this memristor model to build an asymmetric memristive Hopfield neural network (AMHNN), the chaotic dynamics of the proposed AMHNN are investigated and analyzed using fundamental dynamics techniques such as equilibrium stability, bifurcation diagrams, and Lyapunov exponents. According to the findings of this study, the proposed AMHNN possesses a number of complex dynamic properties, including scaling amplitude chaos with coupling strength control, and coexisting uncommon chaotic attractors with initial control and coupling strength control. Significantly, the proposed AMHNN has been observed to exhibit the phenomenon of infinitely persisting uncommon chaotic attractors. In the interim, a system for image encryption based on the proposed AMHNN is constructed. By analyzing correlation, information entropy, and key sensitivity, the devised encryption method reveals a number of benefits. The feasibility of the encryption method is validated through field-programmable gate arrays hardware experiments, and the proposed memristor and AMHNN models have been translated into a Simulink model.
作者机构:
[Kang, Xiatao; Wu, Jiaying; Kang, XT; Xiao, Jingying; Yao, Jiayi] Changsha Univ Sci & Technol, Sch Comp & Commun Engn, Changsha, Peoples R China.;[Wu, Jiaying] Hunan Prov Key Lab Intelligent Proc Big Data Tran, Changsha, Peoples R China.
会议名称:
30th International Conference on Neural Information Processing (ICONIP) of the Asia-Pacific-Neural-Network-Society (APNNS)
会议时间:
NOV 20-23, 2023
会议地点:
Changsha, PEOPLES R CHINA
会议主办单位:
[Wu, Jiaying;Kang, Xiatao;Xiao, Jingying;Yao, Jiayi] Changsha Univ Sci & Technol, Sch Comp & Commun Engn, Changsha, Peoples R China.^[Wu, Jiaying] Hunan Prov Key Lab Intelligent Proc Big Data Tran, Changsha, Peoples R China.
会议论文集名称:
Lecture Notes in Computer Science
关键词:
Deep Learning;Network Pruning;Pruning Before Training;Single-shot Pruning
摘要:
Network pruning prior to training makes generalization more challenging than ever, while recent studies mainly focus on the trainability of the pruned networks in isolation. This paper explores a new perspective on loss implicit decrease of the data to be trained caused by one-batch training during each round, whose first-order approximation we term gradient coupled flow. We thus present a criterion sensitive to gradient coupled flow (GCS), which is hypothesized to capture those weights most sensitive to performance boosting at initialization. Interestingly, our explorations show there exists a linear correlation between generalization and implicit loss decrease based measurements on previous works as well as GCS, which ideally describes causes of accuracy fluctuation in a fine-grained manner. Our code is made public at: https://github.com/kangxiatao/pruning_before_training.
摘要:
BACKGROUND: Dynamic spatial functional network connectivity (dsFNC) has shown advantages in detecting functional alterations impacted by mental disorders using magnitude-only fMRI data. However, complete fMRI data are complex-valued with unique and useful phase information. METHODS: We propose dsFNC of spatial source phase (SSP) maps, derived from complex-valued fMRI data (named SSP-dsFNC), to capture the dynamics elicited by the phase. We compute mutual information for connectivity quantification, employ statistical analysis and Markov chains to assess dynamics, ultimately classifying schizophrenia patients (SZs) and healthy controls (HCs) based on connectivity variance and Markov chain state transitions across windows. RESULTS: SSP-dsFNC yielded greater dynamics and more significant HC-SZ differences, due to the use of complete brain information from complex-valued fMRI data. COMPARISON WITH EXISTING METHODS: Compared with magnitude-dsFNC, SSP-dsFNC detected additional and meaningful connections across windows (e.g., for right frontal parietal) and achieved 14.6% higher accuracy for classifying HCs and SZs. CONCLUSIONS: This work provides new evidence about how SSP-dsFNC could be impacted by schizophrenia, and this information could be used to identify potential imaging biomarkers for psychotic diagnosis.
摘要:
Transformer has achieved outstanding performance in many fields such as computer vision benefiting from its powerful and efficient modelling ability and long-range feature extraction capability complementary to convolution. However, on the one hand, the lack of CNN's innate inductive biases, such as translation invariance and local sensitivity, makes Transformer require more data for learning. On the other hand, labelled hyperspectral samples are scarce due to the time-consuming and costly annotation task. To this end, we propose a semi-supervised hierarchical Transformer model for HSI classification to improve the classification performance of the Transformer with limited labelled samples. In order to perturb the samples more fully and extensively to improve the model performance, two different data augmentation methods are used to perturb the unlabelled samples, and two sets of augmented samples are obtained respectively. The pseudo-label obtained on the original unlabelled sample is used to simultaneously supervise the augmented sample obtained on this unlabelled sample. Among them, only the pseudo-labels above the threshold are retained. To further improve the model stability and classification accuracy, hierarchical patch embedding is proposed to eliminate the mutual interference between pixels. Extensive experiments on three well-known hyperspectral datasets validate the effectiveness of the proposed semi-supervised Transformer model. The experiments show that the model achieves excellent classification accuracy even when there are only 10 labelled samples in each category, which can effectively improve the classification performance of Transformer under small-scale labelled samples.
期刊:
IEEE Transactions on Multimedia,2024年26:2395-2407 ISSN:1520-9210
作者机构:
[Fei Peng] Institute of Artificial Intelligence and Blockchain, Guangzhou University, Guangzhou, China;[Min Long] School of Computer and Communication Engineering, Changsha University of Science and Technology, Changsha, China;[Gangyang Hou; Bo Ou] College of Computer Science and Electronic Engineering, Hunan University, Changsha, China
摘要:
Reversible data hiding in encrypted domain (RDH-ED) can perform data encryption to fulfill the privacy protection of original media and embed additional data for covert communication or access control. However, current researches are focusing on the encrypted images, and little attention is paid to encrypted three-dimensional (3D) models. In this article, a high capacity separable RDH-ED method for encrypted 3D models is proposed based on octree spatial subdivision and multiple most significant bit (multi-MSB) prediction. Firstly, a 3D model is adaptively subdivided into non-overlapping subblocks by octree spatial subdivision, and the vertices in a subblock are classified into embedding set and reference set. To better utilize the spatial correlation of the two sets, the multi-MSB prediction error of the embedding set is used to embed the additional data, and the reference set is used to losslessly recover the embedded set. Then, the model is encrypted by a specified encrypted algorithm. At last, additional data is embedded into the reserved embedding room by multi-MSB substitution. Experimental results show that the proposed method can achieve a higher embedding capacity compared with the state-of-the-art methods, and guarantee the lossless recovery of the 3D model.
期刊:
Computer Methods in Applied Mechanics and Engineering,2024年424:116901 ISSN:0045-7825
通讯作者:
Xiaofeng Yang
作者机构:
[Yunqing Huang] National Center for Applied Mathematics in Hunan, Xiangtan University, 411105, China;School of Computer and Communication Engineering, Changsha University of Science & Technology, Changsha, 410114, China;[Timon Rabczuk] Institute of Structural Mechanics, Bauhaus Universität-Weimar, Weimar 99423, Germany;[Xiaofeng Yang] Department of Mathematics, University of South Carolina, Columbia, SC 29208, USA;[Qing Pan] National Center for Applied Mathematics in Hunan, Xiangtan University, 411105, China<&wdkj&>School of Computer and Communication Engineering, Changsha University of Science & Technology, Changsha, 410114, China
通讯机构:
[Xiaofeng Yang] D;Department of Mathematics, University of South Carolina, Columbia, SC 29208, USA
摘要:
We develop an accurate and robust numerical scheme for solving the incompressible hydrodynamically coupled Cahn–Hilliard system of the two-phase fluid flow system on complex surfaces. Our algorithm leverages a number of efficient techniques, including the subdivision-based isogeometric analysis (IGA) method for spatial discretization, the explicit Invariant Energy Quadratization (EIEQ) method for linearizing nonlinear potentials, the Zero-Energy-Contribution (ZEC) method for decoupling, and the projection method for the Navier–Stokes equation to facilitate fully decoupled type implementations. The integration of these methodologies results in a fully discrete scheme with desired properties such as linearity, second-order temporal accuracy, full decoupling, and unconditional energy stability. The implementation of the scheme is straightforward, requiring the solution of a few elliptic equations with constant coefficients at each time step. The rigorous stability proof of unconditional energy stability and the implementation procedure are given in detail. Numerous numerical simulations on complex curved surfaces are carried out to verify the effectiveness of the proposed numerical scheme.
关键词:
Mobile edge computing;Mobility prediction;Relay assistance;Computing offloading;Unequal splitting of tasks
摘要:
Internet of Vehicles (IoV) is paving the road for the new generation of Intelligent Transportation Systems (ITS), and Mobile Edge Computing (MEC) is enabling IoV to efficiently handle the computation -intensive and time -sensitive tasks. However, this has introduced new challenges such as maximizing computing resources, allocating resources fairly for multi -source tasks concurrently, and dividing tasks for parallelly processing to minimize the latency. To face these challenges, a three-dimensional road vehicle mobility model is constructed, and the problem of offloading strategy and resource allocation among multiple vehicles served by one Road Side Unit (RSU) is investigates to minimize the average latency of multi -source tasks while satisfying the quality of service requirements. To address the Non -deterministic Polynomial -time hardness (NP -hardness) of the problem, we design a Relay -Assisted Parallel Offloading (RAPO) strategy to obtain the optimization solution. Extensive experimental results show that the RAPO strategy introducing relay -assisted nodes can enhance performance in poor scenarios and ensure low -latency multi -tasking under various conditions, especially reducing latency by 39% compared to local computing.
摘要:
Credit fraud brings billions of dollars to banks every year. However, the existing research on fraud detection methods has encountered bottlenecks. Traditional fraud detection models are difficult to improve performance and accuracy in detecting fraudulent transactions. The main reason is that the bank credit dataset is very distorted, and the proportion of positive and negative samples is seriously unbalanced, and secondly, due to data privacy and security issues, the dataset is usually not allowed to be shared between different users. In this paper, we improve and enhance the existing credit card fraud detection system, and propose Approx-SMOTE federated learning credit card fraud detection system (AFLCS), through the method of salting and interference items in the data balancing and learning module, design a federal credit fraud detection algorithm for the data after data balance, and improve the network structure of CNN and learning module, the AFLCS can not only improve the processing time by nearly 30 times without affecting the performance, and even better the privacy and security of the information system, but also reserve space for the subsequent application of the system to data expansion between different banks. We achieved our design on the pysyft framework and tested it with an approved data set. The experiments show that our indicators are higher than the existing solutions.
作者机构:
[Pan, Qing; Zhang, Jin] Changsha Univ Sci & Technol, Sch Comp & Commun Engn, Changsha 410114, Peoples R China.;[Chen, Chong] Chinese Acad Sci, Acad Math & Syst Sci, LSEC, ICMSEC, Beijing 100190, Peoples R China.;[Rabczuk, Timon] Bauhaus Univ Weimar, Inst Struct Mech, D-99423 Weimar, Germany.;[Yang, Xiaofeng; Yang, XF] Univ South Carolina, Dept Math, Columbia, SC 29208 USA.
通讯机构:
[Yang, XF ] U;Univ South Carolina, Dept Math, Columbia, SC 29208 USA.
关键词:
Loop subdivision;IGA-EIEQ;Decoupled;Unconditional energy stability;Phase-field model;Homopolymer blends
摘要:
In this paper, we construct an IGA-EIEQ coupling scheme to solve the phase-field model of homopoly-mer blends on complex subdivision surfaces, in which the total free energy contains a gradient entropy with a concentration-dependent de-Gennes type coefficient and a non-linear logarithmic Flory-Huggins type potential. Based on the EIEQ method, we develop a fully-discrete numerical scheme with the superior properties of linearity, unconditional energy stability, and second-order time accuracy. All we need to do with this fourth-order system is to solve some constant-coefficient elliptic equations by applying a new nonlocal splitting techniqueWe then provide detailed proof of the unconditional energy stability and the practical implementation process. Subdivision approaches show a robust and elegant description of the models with arbitrary topology. Subdivision basis functions serve to define the geometry of the models and represent the numerical solutions. Subdivision-based IGA approach provides us with a good candidate for solving the phase-field model on complex surfaces. We successfully demonstrate the unity of employing subdivision basis functions to describe the geometry and simulate the dynamical behaviors of the phase-field models on surfaces with arbitrary topology. This coupling strategy combining the subdivision-based IGA method and the EIEQ method could be extended to a lot of gradient flow models with complex nonlinearities on complex surfaces.& COPY; 2023 Elsevier Ltd. All rights reserved.