Xidian University

Although machine Learning has demonstrated exceptional applicability in thermographic inspection, precise defect reconstruction is still challenging, especially for complex defect profiles with limited defect sample diversity. Thus, this paper proposes a self-enhancement defect reconstruction technique based on cycle-consistent generative adversarial network (Cycle-GAN) that accurately characterises complex defect profiles and generates reliable artificial thermal images for dataset augmentation , enhancing defect characterisation. By using a synthetic dataset from simulation and experiments, the network overcomes the limited samples problem by learning the diversity of complex defects from finite element modelling and obtaining the thermography uncertainty patterns from practical experiments. Then, an iterative strategy with a self-enhancement capability optimises the characterisation accuracy and data generation performance. The designed loss function structure with cycle consistency and identity loss constrains the GAN's transfer variation to guarantee augmented data quality and defect reconstruction accuracy simultaneously, while the self-enhancement results significantly improve accuracy in thermal images and defect profile reconstruction. The experimental results demonstrate the feasibility of the proposed method by attaining high accuracy with optimal loss norm for defect profile reconstruction with a Recall score over 0.92. The scalability investigation of different materials and defect types is also discussed, highlighting its capability for diverse thermography quantification and automated inspection scenarios. A cyclic self-enhancement technique for complex defect profile reconstruction based on thermographic evaluation, Acta Mech. Sin. 41, 424076 (2025), https://doi.

Metals have traditionally served as the primary functional phase in the development of metamaterials exhibiting epsilon-near-zero (ENZ) and epsilon-negative (EN) responses, albeit with persisting ambiguities regarding their response mechanisms. This paper presents the tunable ENZ (ε′ ~ 0) and EN (ε′ < 0) parameters at the 20-MHz to 1-GHz region based on Cu/CaCu3Ti4O12 (Cu/CCTO) metacomposites. By means of first-principles calculations and multi-physics simulations, the underlying mechanisms governing ENZ and EN responses are unveiled. The intricate pathways through which metacomposites achieve 2 dielectric response mechanisms are delineated: At low Cu content, a weak EN response (|ε′| < 200) was excited by electric dipole resonance, accompanied by ENZ effect; conversely, at high Cu content, due to the increase in effective electron concentration, plasmonic oscillation behavior occurs in the constructed 3-dimensional Cu network, resulting in strong EN response (|ε′| ~ 1,000) in the radio frequency band. These phenomena are explicated through 2 distinct Cu/CCTO models: Cu in an isolated state and a connected network state. This study not only comprehensively elucidates the 2 EN response mechanisms achieved by typical metacomposites with metals as functional phases but also delves into their associated electromagnetic shielding and thermal properties, providing a theoretical basis for their practical applications.

Distant object detection is a difficult problem in LiDAR-based 3D object detection. In recent years, the 3D detection of distant objects has achieved great success with the proposed fusion method of the virtual points generated by depth completion and LiDAR points. However, the inaccuracy of depth completion brings a lot of noise which significantly reduces the detection accuracy. To reduce noise and improve the detection accuracy of distant objects, we propose a solution called VirInteraction, which is a semantic-guided Virtual-LiDAR fusion method to enhance the interaction of virtual points and LiDAR points. Specifically, VirInteraction mainly includes three new designs: 1) Foreground-based adaptive Voxel Denoising (FgVD), 2) Semantic neighboring Sampling (Se-Sampling), and 3) Multi-scale Density-aware Cross Attention (MDC-Attention). FgVD uses Kernel Density Estimation (KDE) to adaptively denoise the foreground and background voxels. Se-Sampling completes the shape cues of distant objects using bidirectional sampling based on self-attention mechanism. Meanwhile, we built on these two designs and VirConvNet to develop a more robust VirInterNet as our virtual-point-based backbone. Finally, MDC-Attention elegantly aggregates the features of the images and points at the feature level according to the density distribution. Extensive experiments on KITTI and nuScenes datasets demonstrate the effectiveness of VirInteraction.

A reconfigurable holographic metasurface (HM) with multifunctional modulation of radiation and scattering for conformal applications is designed in this paper. Based on optical holography theory, a holographic conformal modulation mechanism is proposed, and the conformal surface impedance distribution of HM is derived. To illustrate this mechanism, the designed conformal reconfigurable HM is used to demonstrate a series of radiation and scattering modulation functions, with its reconfigurable property enabling dynamic beam control. In radiation mode, beam scanning with wide angle from −50° to 50° is achieved. In scattering mode, specific responses are generated under different incident angles, including beam steering under oblique incidence within ±60°, multi-beam splitting within ±60° under normal incidence, and diffuse reflection. Low radar cross section (RCS) is exhibited over a wide frequency band from 7.2 to 25 GHz. The designed conformal reconfigurable HM shows high adaptability to cylindrical platforms, insensitivity to oblique incidence, and stability in beam deflection angles, which provides an innovative technical approach for information transmission and stealth in conformal devices.

In non-cooperative scenarios, signal modulation style analysis has received considerable interest as a method for avoiding malicious attacks in modern cognitive communication systems. However, existing signal clustering methods often overlook the potential of the spectrum and fail to address the issue of local optima commonly encountered in the field of signal clustering. In this paper, we introduce a novel approach to accurately cluster modulation styles, termed Spectrum Augmentation for Contrastive Clustering (SACC). SACC proposes two main components, namely, spectrum augmentation and self-labeling optimization, building upon the foundation of contrastive clustering. Specifically, spectrum augmentation (SA) is employed in the first stage to facilitate effective deep semantic feature extraction, leveraging contrastive learning. SA is introduced as a signal spectrum-based data augmentation method, exploiting the temporal and frequency representation capability of signals. In the second stage, a reliable negative self-labeling optimization method is proposed atop deep clustering to address the issue of local optima arising from the lack of label guidance in signal clustering. Extensive experimental results validate the effectiveness of the SACC method, SACC achieving significant performance improvements with a straightforward design. Specifically, SACC achieves an accuracy rate of 80.1% and a normalized mutual information score of 0.826 on two publicly available datasets, demonstrating the superiority of our proposed approach.

A new zero space orthogonal projection (ZSOP) method is presented to synthesize the antenna array pattern. The expected pattern vector is divided into two parts, the main lobe vector and the side lobe vector. In order to maximize the side lobe attenuation, the optimal solution of the antenna element excitation vector should be placed in the zero space of the side lobe steering matrix. Therefore, the optimal solution of the antenna element excitation vector must be in the orthogonal projection space of the conjugate transpose matrix of the side lobe steering matrix. Thus, the pattern synthesis equation can be transformed into a new form. The solution of the new equation can ensure that the optimal solution of the antenna element excitation vector can form a pattern with an extremely low null beam level. Examples are used to demonstrate the advantages of the new method. Simulation results show that the new method can form patterns with excellent null beam levels that far exceed the performance of other methods. The new method can work with non-iterative calculation steps. It requires only slightly more computation amount than the traditional least square method to complete the pattern synthesis task.

Due to its simplicity in circuit implementation, conventional Data Weighted Averaging (DWA) is commonly used to calibrate capacitor mismatch in feedback DACs. However, for low-amplitude input signal, the non-random use of capacitor elements causes periodic mismatch errors, leading to harmonic distortion in the signal band. Consequently, this results in significant harmonic distortion within the signal band. This paper proposes a randomized DWA algorithm that utilizes the amplitude of the input signal to control the starting position of DAC elements for each cycle, thereby suppressing tones caused by DAC element mismatches. Compared to the conventional DWA algorithm, this approach achieves higher linearity while reducing DAC switching activities. To evaluate the proposed algorithm, a second-order discrete-time sigma-delta modulator model was designed. Simulation results indicate that the proposed algorithm achieves up to a 14% reduction in switching activities and extending the dynamic range by approximately 5 dB compared to the conventional randomized DWA algorithm.

In today's information age, table images play a crucial role in storing structured information, making table image recognition technology an essential component in many fields. However, accurately recognizing the structure and text content of various complex table images has remained a challenge. Recently, large language models (LLMs) have demonstrated exceptional capabilities in various natural language processing tasks. Therefore, applying LLMs to the correction tasks of structure and text content after table image recognition presents a novel solution. This paper introduces a new method, TableGPT, which combines table recognition with LLMs and develops a specialized multimodal agent to enhance the effectiveness of table image recognition. Our approach is divided into four stages. In the first stage, TableGPT_agent initially evaluates whether the input is a table image and, upon confirmation, uses algorithms such as the transformer for preliminary recognition. In the second stage, the agent converts the recognition results into HTML format and autonomously assesses whether corrections are needed. If corrections are needed, the data are input into a trained LLM to achieve more accurate table recognition and optimization. In the third stage, the agent evaluates user satisfaction through feedback and applies superresolution algorithms to low-quality images, as this is often the main reason for user dissatisfaction. Finally, the agent inputs both the enhanced and original images into the trained model, integrating the information to obtain the optimal table text representation. Our research shows that trained LLMs can effectively interpret table images, improving the Tree Edit Distance Similarity (TEDS) score by an average of 4% even when based on the best current table recognition methods, across both public and private datasets. They also demonstrate better performance in correcting structural and textual errors. We also explore the impact of image superresolution technology on low-quality table images. Combined with the LLMs, our TEDS score significantly increased by 54%, greatly enhancing the recognition performance. Finally, by leveraging agent technology, our multimodal model improved table recognition performance, with the TEDS score of TableGPT_agent surpassing that of GPT-4 by 34%.

Anti-ambipolar transistors (AAT) are considered as a breakthrough technology in the field of electronics and optoelectronics, which is not only widely used in diverse logic circuits, but also crucial for the realization of high-performance photodetectors. The anti-ambipolar characteristics arising from the gate-tunable energy band structure can produce high-performance photodetection at different gate voltages. As a result, this places higher demands on the parametric driving range (ΔVg) and peak-to-valley ratio (PVR) of the AAT. Here, we demonstrate a high-performance photodetector with anti-ambipolar properties based on a van der Waals heterojunction of MoTe2/MoS2. Flexible modulation of carrier concentration and transport by gate voltage achieves a driving voltage range ΔVg as high as 38.4 V and a peak-to-valley ratio PVR of 1.6 × 102. Most importantly, MoTe2/MoS2 exhibits a pronounced gate-tunable photoresponse, which is attributed to the modulation of photogenerated carrier transport by gate voltage. The MoTe2/MoS2 heterojunction photodetector exhibits excellent performance, including an impressive responsivity of 17 A/W, a high detectivity of 4.2 × 1011 cm Hz1/2 W−1, an elevated external quantum efficiency of 4 × 103 %, and a fast response time of 21 ms. Gate-tunable photodetectors based on MoTe2/MoS2 heterostructures AAT have potential to realize optoelectronic devices with high performance, providing a novel strategy to achieve high-performance photodetection.

The time-multiplexing super-resolution concept requires post-processing for extracting the super-resolved image. Moreover, to perform the post-processing image restoration, one needs to know the exact high-resolution encoding pattern. Both of these limiting conditions are overcome by the method and experiment reported in this letter.

Combining perovskite oxides (BaTiO3 and SrTiO3) with indium-gallium-zinc-oxide (IGZO) has great potential for developing thin film transistors (TFT) due to the ferroelectricity, extreme permittivity and promotion for gate-controlled ability and surface passivation. In this work, the heterojunction of BaTiO3/IGZO and SrTiO3/IGZO were prepared on sapphire by magnetron sputtering. The surface morphologies, crystalline structures, chemical compositions, and the band alignments of the deposited films and related heterojunctions were investigated. The BaTiO3, SrTiO3, IGZO films exhibited a smooth surface, decent film quality, and low oxygen vacancies. The valence band offset (ΔEv) of BaTiO3/IGZO, SrTiO3/IGZO was determined to be 0.22 ± 0.03 eV, 0.16 ± 0.05 eV, respectively, using the Ga 2p3/2, Zn 2p3/2, and In 3d5/2 energy levels as references. It was found that BaTiO3/IGZO form a straddling type I alignment with a conduction band offset (ΔEc) of 0.17 ± 0.03 eV, and SrTiO3/IGZO form a staggered type II alignment with a ΔEc of −0.36 ± 0.04 eV. These results demonstrate that the feasible formation of BaTiO3/IGZO and SrTiO3/IGZO heterojunctions with smooth surface and decent quality, and BaTiO3 could play important role in surface passivation and electron confinement for IGZO TFTs, which is important for design IGZO/ferroelectric heterojunction multifunctional devices.

This study presents the design and analysis of a thermal energy harvester that integrates a thermoacoustic engine (TAE) with a honeycomb-structured triboelectric nanogenerator (H-TENG), referred to as TAEH-TENG. This design is specifically developed to demonstrate the potential of thermal energy harvesting for low-power Internet of Things (IoT) applications. By leveraging the high energy conversion efficiency of TAEs and the exceptional robustness of H-TENGs, this harvester overcomes the limitations of traditional designs, which often involve complex or costly components. The experimental results revealed the oscillation characteristics of the TAEH-TENG: by utilizing a hot heat exchanger (HHE) with a length of 10 cm, the system can sustain oscillation over 150–350 °C. Furthermore, the harvester is capable of generating an open-circuit voltage of 25 V, an RMS current of 0.98 μA, and a peak power output of 0.48 mW, representing the highest power output achieved to date in comparison to previous studies. To further showcase the harvester's capability, an ultra-low-power IoT node was developed. Solely powered by the TAEH-TENG, the IoT node achieved cold-start, conducted in situ temperature measurement five times, and transmitted the data via Bluetooth within 120 s. This study not only showcases a fully self-powered IoT application but, more importantly, significantly advances the technology beyond the previous limitations faced by thermoacoustic and triboelectric integrations. By demonstrating the capability to power an ultra-low-power IoT node, this research highlights the TAEH-TENG's potential for practical, real-world energy solutions, marking a significant milestone in the deployment of heat-powered IoT applications.

Federated Learning (FL) is a technique that can learn a global machine-learning model at a central server by aggregating locally trained models. This distributed machine-learning approach preserves the privacy of local models. However, FL systems are inherently vulnerable to significant security challenges such as cyber-attacks, handling non-independent and identically distributed (non-IID) data, and data privacy concerns. This systematic literature review addresses these issues by examining advanced neural network models, feature engineering methods, and privacy-preserving techniques within intrusion detection systems (IDS) for FL environments. These are key elements for improving the security of FL systems. To the best of our knowledge, this review is among the first to comprehensively explore the combined impacts of these technologies. We analyzed 88 studies published between 2021 and October 2024. This study offers valuable insights for future research directions, including scaling FL in a real-world environment.

Formal context restoration is a recently developing topic in the field of formal concept analysis (FCA). Its goal is to restore a formal context from some known formal concepts. Each type of formal concept uses its unique perspective to restore formal contexts or concept lattices. This paper investigates the relationship among five types of basic concepts: the object concepts, the attribute concepts, the join-irreducible concepts, the meet-irreducible concepts, and the formal concepts in concept reducts. This paper first studies the elementary relationship among the basic sets of formal concepts and presents these relationship in a Venn diagram. Then the relationship among the basic concept sets are explored from a restoration perspective, specifically including the relationship among basic concepts from the perspectives of formal context restoration and concept lattice restoration. These relationship are then used to study the transformation between the basic concept sets, and are summarised in a formal context and its concept lattice. Finally, a practical case is given to illustrate the relationship among the basic concept sets explored in this paper, including the elementary relationship, and the relationship from the perspectives of formal context restoration and concept lattice restoration.

In the digital age, 5G technology has significantly enhanced global communication capabilities, providing strong support for various industries. However, 5G falls short of achieving comprehensive intelligence within the Internet of Things (IoT), propelling the development of 6G technology. 6G is expected to further enhance network performance by integrating advanced technologies such as AI and quantum communication, while expanding application scenarios to include communication in extreme environments for comprehensive global connectivity. This paper proposes an innovative fronthaul architecture that applies Quantum Key Distribution (QKD) technology to the 5G fronthaul, leveraging its unconditionally secure key generation and distribution mechanism. In this design, costlier Alice devices are placed on the AAU side, while less expensive Bob devices are positioned on the DU side, optimizing deployment to reduce engineering costs and lower deployment barriers. The feasibility of this architecture is demonstrated by calculating the secure key rate. Comparative analysis with existing research is performed to clarify future research directions. These innovations not only enhance 5G network security but also offer new solutions for the security requirements of 6G networks, such as data security, network resilience, and algorithm transparency. Our research offers strategic value for future network security, laying the groundwork for reliable networks.

Let $\{S_{n}, n\geq 0\}$ be a random walk whose increment distribution belongs without centering to the domain of attraction of an $\alpha$-stable law, that is, there are scaling constants $a_{n}$ such that the sequence $S_{n}/a_{n}$, n=1,2,…, converges weakly, as $n\to\infty$, to a random variable having an $\alpha$-stable distribution. Let $S_{0}=0$, L_{n}:=\min (S_{1},…,S_{n})\quadand\quad\tau_{n}:=\min \{ 0\leq k\leq n\colon S_{k}=\min (0,L_{n})\}. Assuming that $S_{n}\leq h(n)$, where h(n) is $o(a_{n})$ as $n\to\infty$ and the limit $\lim_{n\to\infty}h(n)\in [-\infty,+\infty]$ exists, we prove several limit theorems describing the asymptotic behaviour of the functionals $\mathbf{E}[ e^{\lambda S_{\tau_{n}}}; S_{n}\leq h(n)], \qquad \lambda>0,$ as $n\to\infty$. The results obtained are applied to study the survival probability of a critical branching process evolving in an extremely unfavourable random environment. Bibliography: 15 titles.

The biological photoreceptors in the retina convert light information into spikes, inspiring the emergence of artificial photoelectric spiking neurons. However, due to the lack of biocompatible and biodegradable characteristics, artificial photoelectric spiking neurons based on threshold switching (TS) devices are not available for bio‐integrated optical medical diagnostics and neuromorphic computing. Here, an artificial photoelectric spiking neuron integrated with a physically transient memristor and photodetector for UV perception is proposed. The transient memristor with a MgO:Mg resistive layer implemented by the co‐sputtering process of MgO and Mg targets shows highly robust TS performance, while the ZnO‐based transient photodetector can selectively detect UV light at power densities below 10 mW cm⁻². More interestingly, the frequency of the firing spikes generated by artificial photoelectric spiking neuron increases with the enhancement of UV light intensity. In addition, the recognition accuracy of UV information extracted from the surrounding environment reaches ≈99.8% by spiking neural network consisting of photoelectric spiking neuron when the object that blended into the background are not easily detected. This work demonstrates that the functions of the biological photoreceptors may be truly mimicked by artificial photoelectric spiking neuron with transiency, expanding its application in optical disease diagnosis and implantable visual neuromorphic computing.

The derivative mapping plays a central role in many applications, including, most noticeably, differential cryptanalysis (Biham and Shamir in J Cryptol 4:3–72, 1991; Nyberg, in: Advances in Cryptology EUROCRYPT’93, Lecture Notes in Computer Science, 1994). The ambiguity and deficiency of functions measure how close the derivative mapping is to be the injection and surjection, respectively. In this paper, by studying some special linearized polynomials, we determine the exact values of ambiguity, deficiency and differential uniformity of some permutation trinomials over $\mathbb {F}_{2^n}$.