Large language models (LLMs), such as ChatGPT released by OpenAI, have attracted significant attention from both industry and academia due to their demonstrated ability to generate high-quality content for various tasks. Despite the impressive capabilities of LLMs, there are growing concerns regarding their potential risks in various fields, such as news, education, and software engineering. Recently, several commercial and open source LLM-generated content detectors have been proposed, which, however, are primarily designed for detecting natural language content without considering the specific characteristics of program code. This article aims to fill this gap by proposing a novel ChatGPT-generated code detector, CodeGPTSensor, based on a contrastive learning framework and a semantic encoder built with UniXcoder. To assess the effectiveness of CodeGPTSensor on differentiating ChatGPT-generated code from human-written code, we first curate a large-scale Human and Machine comparison Corpus (HMCorp), which includes 550k pairs of human-written and ChatGPT-generated code (i.e., 288k Python code pairs and 222k Java code pairs). Based on the HMCorp dataset, our qualitative and quantitative analysis of the characteristics of ChatGPT-generated code reveals the challenge and opportunity of distinguishing ChatGPT-generated code from human-written code with their representative features. Our experimental results indicate that CodeGPTSensor can effectively identify ChatGPT-generated code, outperforming all selected baselines.

Automatic polyp segmentation from colonoscopy videos is a critical task for the development of computer-aided screening and diagnosis systems. However, accurate and real-time video polyp segmentation (VPS) is a very challenging task due to low contrast between background and polyps and frame-to-frame dramatic variations in colonoscopy videos. We propose a novel embedding-unleashing framework consisting of a proposal-generative network (PGN) and an appearance-embedding network (AEN) to comprehensively address these challenges. Our framework, for the first time, models VPS as an appearance-level semantic embedding process to facilitate generate more global information to counteract background disturbances and dramatic variations. Specifically, PGN is a video segmentation network to obtain segmentation mask proposals, while AEN is a network we specially designed to produce appearance-level embedding semantics for PGN, thereby unleashing the capability of PGN in VPS. Our AEN consists of a cross-scale region linking (CRL) module and a cross-wise scale alignment (CSA) module. The former screens reliable background information against background disturbances by constructing linking of region semantics, while the latter performs the scale alignment to resist dramatic variations by modeling the center-perceived motion dependence with a cross-wise manner. We further introduce a parameter-free semantic interaction to embed the semantics of AEN into PGN to obtain the segmentation results. Extensive experiments on CVC-612 and SUN-SEG demonstrate that our approach achieves better performance than other state-of-the-art methods. Codes are available at https://github.com/zhixue-fang/EUVPS.

With the widespread use of unmanned aerial vehicles (UAVs), the issue of data confidentiality is becoming more and more prominent. For this reason, this paper intends to build a UAV-enabled virtual antenna array (UVAA) and communicate with the BS on a vessel in a maritime environment using cooperative beamforming (CB) techniques. To improve safety, the UAV elements can carry optimal excitation current weights and fly to appropriate locations for CB transmission. However, this will result in more energy consumption. To address several critical issues in UAV communication, a secure communication multi-objective optimization problem (SCMOP) is proposed to simultaneously improve the total secrecy rate, the total maximum sidelobe levels (SLLs), and the total motion energy consumption of the UAVs by jointly optimizing the position and the excitation current weights. Because the SCMOP is non-convexity and NP-hard, we adopt a non-dominated sorting whale optimization algorithm(INSWOA) with chaotic solution initialization, optimal position update based on the sine cosine algorithm (SCA), and adaptive weights to solve the problem. Experiments show that this method can better solve the SCMOP and is superior to some existing standard methods.

Automatic polyp segmentation from colonoscopy images is an essential prerequisite for the development of computer-assisted therapy. However, the complex semantic information and the blurred edges of polyps make segmentation extremely difficult. In this paper, we propose a novel semi-supervised polyp segmentation framework using affinity contrastive learning (ACL-Net), which is implemented between student and teacher networks to consistently refine the pseudo-labels for semi-supervised polyp segmentation. By aligning the affinity maps between the two branches, a better polyp region activation can be obtained to fully exploit the appearance-level context encoded in the feature maps, thereby improving the capability of capturing not only global localization and shape context, but also the local textural and boundary details. By utilizing the rich inter-image affinity context and establishing a global affinity context based on the memory bank, a cross-image affinity aggregation (CAA) module is also implemented to further refine the affinity aggregation between the two branches. By continuously and adaptively refining pseudo-labels with optimized affinity, we can improve the semi-supervised polyp segmentation based on the mutually reinforced knowledge interaction among contrastive learning and consistency learning iterations. Extensive experiments on five benchmark datasets, including Kvasir-SEG, CVC-ClinicDB, CVC-300, CVC-ColonDB and ETIS, demonstrate the effectiveness and superiority of our method. Codes are available at https://github.com/xiewende/ACL-Net.

Software vulnerabilities damage the functionality of software systems. Recently, many deep learning-based approaches have been proposed to detect vulnerabilities at the function level by using one or a few different modalities (e.g., text representation, graph-based representation) of the function and have achieved promising performance. However, some of these existing studies have not completely leveraged these diverse modalities, particularly the underutilized image modality, and the others using images to represent functions for vulnerability detection have not made adequate use of the significant graph structure underlying the images. In this paper, we propose MVulD, a multi-modal-based function-level vulnerability detection approach, which utilizes multi-modal features of the function (i.e., text representation, graph representation, and image representation) to detect vulnerabilities. Specifically, MVulD utilizes a pre-trained model (i.e., UniXcoder) to learn the semantic information of the textual source code, employs the graph neural network to distill graph-based representation, and makes use of computer vision techniques to obtain the image representation while retaining the graph structure of the function. We conducted a large-scale experiment on 25,816 functions. The experimental results show that MVulD improves four state-of-the-art baselines by 30.8%-81.3%, 12.8%-27.4%, 48.8%-115%, and 22.9%-141% in terms of F1-score, Accuracy, Precision, and PR-AUC respectively.

With the rapid development of wireless communication, unmanned aerial vehicle (UAV) networks have received extensive attention and been applied in many fields, but some challenges exist in their applications, such as the issue of security when implementing communication. In this paper, a virtual antenna array (VAA) is formed in multiple UAV units using collaborative beamforming (CB) technology. Under the interference of multiple eavesdroppers, secure communication with the ground base station (BS) is achieved. To achieve better performance, we formulate a multi-objective optimization problem for UAV network security communication (SCMOP), and jointly optimize the positions and excitation current weights of UAVs to set the null values in the direction of known eavesdroppers, reduce the sidelobe levels (SLLs) and enhance the directivity of the main lobe (ML). Since the formulated SCMOP is an NP-hard problem, we propose an improved the third non-dominated sorting genetic algorithm (IMNSGA-III) with chaos operator and crossover and mutation operators to solve the problem in this paper. The simulation results show that the IMNSGA-III can solve the SCMOP well and obtain the best results compared with other benchmark algorithms.

Correct and robust translational motion compensation is necessary but challenging for inverse synthetic aperture radar (ISAR) imaging and target recognition. On the one hand, parametric translational motion compensation methods only effective for polynomial translational models, on the other hand, the noise robustness of range alignment-autofocus methods is not satisfactory. Therefore, this work proposes a joint range alignment and autofocus method based on Combined Broyden-Fletcher-Goldfarb-Shanno algorithm and whale optimization algorithm (BFGS-WOA). The method aims to achieve accurate and robust compensation of profile shift and the phase error simultaneously without relying on a specific translational model. Specifically, we use Laplacian entropy and squared envelope entropy to construct a dynamic objective function. It can improve the robustness of translational error estimation. The translational error of each pulse can be estimated by minimizing the objective function. We adopt BFGS to solve the optimization to ensure the global convergence. WOA is introduced to determine the optimal step size of the iteration. Finally, the estimated translational error is used to perform range alignment and autofocus simultaneously. Experimental results of measured datasets demonstrate that the proposed method outperforms existing methods and has strong robustness for low signal to noise ratio (SNR) and sparse aperture.

We present a novel deep network (namely BUSSeg) equipped with both within- and cross-image long-range dependency modeling for automated lesions segmentation from breast ultrasound images, which is a quite daunting task due to (1) the large variation of breast lesions, (2) the ambiguous lesion boundaries, and (3) the existence of speckle noise and artifacts in ultrasound images. Our work is motivated by the fact that most existing methods only focus on modeling the within-image dependencies while neglecting the cross-image dependencies, which are essential for this task under limited training data and noise. We first propose a novel cross-image dependency module (CDM) with a cross-image contextual modeling scheme and a cross-image dependency loss (CDL) to capture more consistent feature expression and alleviate noise interference. Compared with existing cross-image methods, the proposed CDM has two merits. First, we utilize more complete spatial features instead of commonly used discrete pixel vectors to capture the semantic dependencies between images, mitigating the negative effects of speckle noise and making the acquired features more representative. Second, the proposed CDM includes both intra- and inter-class contextual modeling rather than just extracting homogeneous contextual dependencies. Furthermore, we develop a parallel bi-encoder architecture (PBA) to tame a Transformer and a convolutional neural network to enhance BUSSeg's capability in capturing within-image long-range dependencies and hence offer richer features for CDM. We conducted extensive experiments on two representative public breast ultrasound datasets, and the results demonstrate that the proposed BUSSeg consistently outperforms state-of-the-art approaches in most metrics.

Translational motion compensation for maneuvering targets is a challenging step in inverse synthetic aperture radar (ISAR) imaging. On one hand, nonuniform translational motion and rotational motion of maneuvering targets result in nonlinear bending of the range profile and create high-order slow time phases. On the other hand, radar cross section (RCS) fluctuation of targets may cause a sparse aperture and, therefore, sharply weakens the correlation between the pulses. Under this condition, traditional methods of translational motion compensation are sometimes difficult to get ideal results. Therefore, we propose an effective and novel method to achieve translational motion compensation for maneuvering targets with sparse aperture. In this method, the translational and rotational motions are modeled as cubic polynomial and quadratic polynomial, respectively. Based on this model, we use the echo to construct a 1-D optimization to estimate all translational parameters simultaneously and use the root-mean-square prop-momentum gradient descent (RMSprop-MGD) algorithm to solve the optimization to maintain the accuracy of parameter estimation. Finally, we use those translational parameters to compensate for the translational motion. Since this method converts the multiparameter optimization into a 1-D optimization, we named it dimension compressed optimization (DCO). Experimental results of the measured dataset prove that the proposed method is effective for full-aperture echoes, random sparse echoes, and block sparse echoes. Moreover, this method is robust under a low pulse sampling rate and low signal-to-noise ratio (SNR) environment.

Accurate translational motion compensation is critical for ISAR imaging, especially for datasets with low signal-to-noise ratio (SNR) or sparse aperture. Among the existing translational motion compensation methods, the methods that based on the polynomial model have profound noise robustness but are easily affected by the variation of the reference distance. The methods that based on the range profile alignment have good versatility but are far from ideal under low SNR and sparse aperture environment. Therefore, we propose a novel and robust translational motion compensation method to overcome the shortcoming of the existing methods. Specifically, we estimate the translational phase error sequence by maximizing Laplacian average gray level and peak value. It can improve the robustness of translational motion compensation for noise and sparse aperture. Since such principle creates a large-scale multi-objective optimization problem, we design weighted optimization framework - second-order drift particle swarm optimization (WOF - SDPSO) to solve it accurately. WOF-SDPSO achieves full exploration of high-dimensional variable space by variable dimension reducing, random drift update, and second-order oscillation convergence. These ensure the global optimal solution easier to be found. Experimental results of the measured dataset prove that the proposed method is effective and accurate under the conditions of low SNR and sparse aperture. Moreover, the proposed method is noise robust for polynomial translational motion model with sparse aperture.

Generally, high-resolution imaging for highly squinted synthetic aperture radar (SAR) data is a difficult problem due to large range migration. Thus, when trying to solve this nontrivial problem, an azimuth-variant Doppler will arise, thereby leading to phase errors containing the azimuth spatial-variant (ASV) component. In this article, we analyze the characteristics of highly squinted SAR data and propose a new full-aperture ASV phase error autofocus algorithm. In this new algorithm, the accurate and suitable phase error signal model for highly squinted SAR data is derived. Moreover, the closed-form solution of the relationship between a distorted image and a focused image is also explicitly revealed. Furthermore, in this newly proposed algorithm, an accurate estimation of nonlinear ASV phase error is established based on the maximum contrast of the SAR imagery. In addition, an iterative gradient-based solver is introduced. The advantage of this new method provides a simple yet effective approach while being able to eliminate the ASV phase errors. More importantly, the accuracy of this new method using the full-aperture data is independent of SAR imaging algorithms. As a result, the proposed new method can be easily embedded in many existing imaging algorithms to produce focused imagery. Finally, two real highly squinted SAR data sets are provided to validate the advantages of our algorithm.