Current multimodal misinformation detection (MMD) methods often assume a single source and type of forgery for each sample, which is insufficient for real-world scenarios where multiple forgery sources coexist. The lack of a benchmark for mixed-source misinformation has hindered progress in this field. To address this, we introduce MMFakeBench, the first comprehensive benchmark for mixed-source MMD. MMFakeBench includes 3 critical sources: textual veracity distortion, visual veracity distortion, and cross-modal consistency distortion, along with 12 sub-categories of misinformation forgery types. We further conduct an extensive evaluation of 6 prevalent detection methods and 15 Large Vision-Language Models (LVLMs) on MMFakeBench under a zero-shot setting. The results indicate that current methods struggle under this challenging and realistic mixed-source MMD setting. Additionally, we propose MMD-Agent, a novel approach to integrate the reasoning, action, and tool-use capabilities of LVLM agents, significantly enhancing accuracy and generalization. We believe this study will catalyze future research into more realistic mixed-source multimodal misinformation and provide a fair evaluation of misinformation detection methods.

Quantum Computing (QC) leverages quantum mechanics to manipulate quantum information, holding greater potential than classical computing. To fully exploit QC’s potential, it is crucial to ensure the reliability and quality of quantum programs. Research on quantum program testing is still at its early stage, in which some distinctive features of quantum programs, e.g., superposition and entanglement, may be overlooked, and the fault detection capability and testing effectiveness are rather limited. Besides, the input space of quantum programs may exponentially grow when the number of qubits increases, posing great challenges to testing quantum programs. It is imperative to develop a proper testing strategy to effectively select the potential failure-causing test cases and detect faults faster. In this paper, test cases with both basis states and superposition ones are considered and generated to cover more input space. A hybrid distance measurement method based on quantum fidelity and Hamming distance is presented for measuring the similarity among quantum test cases. Furthermore, a Dynamic Random Testing strategy based on Hybrid distance metrics (DRT-H) for quantum programs is proposed, which combines the hybrid distance metrics and the feedback mechanism of the classical Dynamic Random Testing (DRT) strategy to adjust the testing profile and guide the test case selection. Experimental studies demonstrate that the proposed DRT-H strategy outperforms the baseline testing strategies in most cases.

With the continuous advancement of quantum computing (QC), the demand for high-quality quantum programs (QPs) is growing. To avoid program failure, in software engineering, the technology of automatic program repair (APR) employs appropriate patches to remove potential bugs without the intervention of a human. However, the method tailored for repairing defective QPs is still absent. This article proposes, to the best of our knowledge, a new APR method named UnitAR that can repair QPs via unitary operation automatically. Based on the characteristics of superposition and entanglement in QC, the article constructs an algebraic model and adopts a generate-and-validate approach for the repair procedure. Furthermore, the article presents two schemes that can respectively promote the efficiency of generating patches and guarantee the effectiveness of applying patches. For the purpose of evaluating the proposed method, the article selects 29 mutated versions as well as five real-world buggy programs as the objects and introduces two traditional APR approaches GenProg and TBar as baselines. According to the experiments, UnitAR can fix 23 buggy programs, and this method demonstrates the highest efficiency and effectiveness among three APR approaches. Besides, the experimental results further manifest the crucial roles of two constituents involved in the framework of UnitAR .

In this paper, we delve into semi-supervised 2D human pose estimation. The previous method ignored two problems: (i) When conducting interactive training between large model and lightweight model, the pseudo label of lightweight model will be used to guide large models. (ii) The negative impact of noise pseudo labels on training. Moreover, the labels used for 2D human pose estimation are relatively complex: keypoint category and keypoint position. To solve the problems mentioned above, we propose a semi-supervised 2D human pose estimation framework driven by a position inconsistency pseudo label correction module (SSPCM). We introduce an additional auxiliary teacher and use the pseudo labels generated by the two teacher model in different periods to calculate the inconsistency score and remove outliers. Then, the two teacher models are updated through interactive training, and the student model is updated using the pseudo labels generated by two teachers. To further improve the performance of the student model, we use the semi-supervised Cut-Occlude based on pseudo keypoint perception to generate more hard and effective samples. In addition, we also proposed a new indoor overhead fisheye human keypoint dataset WEPDTOF-Pose. Extensive experiments demonstrate that our method outperforms the previous best semi-supervised 2D human pose estimation method. We will release the code and dataset at https://github.com/hlz0606/SSPCM

Deep neural networks (DNNs) have achieved tremendous development while they may encounter with incorrect behaviors and result in economic losses. Identifying the most represented data become critical for revealing incorrect behaviours and improving the quality DNN-driven systems. Various testing strategies for DNNs have been proposed. However, DNN testing is still at early stage and existing strategies might not sufficiently effective. Dynamic random testing (DRT) strategy uses the feedback mechanism to guide the test case selection, which has been proved to be effective in fault detection. However, its efficacy for Natural Language Processing (NLP) DNN models has not been thoroughly studied. In this paper, a Distance-based DRT with prioritization (D-DRT-P) is proposed, which combines the priority information and distance information into DRT to guide the selection of test cases and testing profile adjustment. Empirical studies demonstrate that D-DRT-P can improve the fault detecting effectiveness than other test prioritization strategies in most cases.