Abstract The spread of disinformation on online social media has caused massive concern. Existing disinformation detection methods neglect the diverse compositional forms of tweets in real-life scenarios, making them less applicable and effective in social media settings. Meanwhile, these methods use pattern cues but overlook important aspects such as syntax, lexicon, and shallow visual semantics, and lack attention to factual content such as time, place, and person relay in both text and images, thus failing to fully explore features of disinformation and limiting detection accuracy. Furthermore, with the popularity of large language models (LLMs), the tweets generated by these models make the style of disinformation more subtle. Since existing datasets are mostly human-generated and lack style diversity, it results in weak detection capabilities of methods trained on these datasets. To address these challenges, a dual-feature adaptive framework for multimodal disinformation detection is proposed. The framework first using a similarity-based algorithm adaptively handles different tweet forms. It then enhances pattern features by bridging multimodal output from single-modal pretrained modal, and factual features are subsequently extracted using a zero-shot method based on a large vision language model. Finally, an expert network aggregates and reweights the dual-feature representation for tweets using an LLM-text detector in gating strategy. This paper also presents two multimodal disinformation datasets that include both LLM-generated and human-generated tweets reflecting real-world scenarios. The true tweets in datasets are diverse in style, while the fake tweets are more misleading. Experimentally verified, the proposed method outperforms baseline methods by an accuracy of 1.04% and 0.72% on typical datasets while also achieving a minimum accuracy drop of 0.65% and 0.87% on the proposed dataset.

In the study of dialogue system, personalized dialogue mainly focuses on the semantic matching degree between response and role. However, various factors, such as semantic style, dialogue noise, and sparse personality information, can affect the performance of the model. For this purpose, we construct a novel personalized dialogue model. Based on the personality description information and conversation history, it uses the information enhancement algorithm to cluster the personality description text into a number of fine sparse categories, and uses the feature classifier to precisely select the features highly relevant to the current dialogue situation according to the input query content. At the same time, the history information selector will fine-filter the conversation history, retaining the parts that are valuable for generating replies. We combine the processed personality description text with the filtered historical context information, and send it to the decoder through the feature cue learning strategy for deep processing to generate personalized responses. Our model integrates the proposed algorithms, classifiers, selectors and feature cue learning strategies to build a complete dialogue system. Experiments on two datasets show that our model is superior to other models in terms of consistency and coherence.

Ultrawideband (UWB)-based localization is widely used in environments inaccessible to global navigation satellite system signals. To improve the precision of UWB-based localization, a robust distributed adaptive extended unbiased finite impulse response (EFIR) filtering algorithm is developed. The algorithm is designed to reduce round-off errors and adaptively adjust noise covariances using the expectation-maximization (EM) approach. Based on extensive experimental testing, the EFIR algorithm is shown to outperform the distributed extended Kalman filter-based algorithm and distributed EFIR filter-based algorithm under harsh conditions.

Open-source code repositories play an important role in software development, but they also introduce a slew of security issues. Firstly, everyone can use open-source projects and libraries from the third-party ecosystem, which increases the risk of vulnerabilities attacking. Secondly, it may cause a domino effect and make these products inherit these vulnerabilities when referring to vulnerable repositories. Traditional technical methods were unable to detect these public flaws in a timely manner, leaving these developers in an insecure situation. Although vulnerability management institutes like CVE and NVD provide inadequate coverage, leading to a lack of timely, reliable, and detailed information about open-source projects' vulnerabilities. To better detect and repair vulnerability, we designed a vulnerability intelligence mining system named DeepVuler based on threads analysis and changed code in open-source communities using machine learning. We choose and define a series of effective features extracted from open-source communities to early infer vulnerability intelligence. Our result shows that two proposed models of DeepVuler achieve a detection rate of 0.979 in threads and 0.890 in changed codes. Besides, the detection from DeepVuler is often days or weeks ahead of official vulnerability disclosure.

At present, the study of upper‐limb posture recognition is still in the primary stage; due to the diversity of the objective environment and the complexity of the human body posture, the upper‐limb posture has no public dataset. In this paper, an upper extremity data acquisition system is designed, with a three‐channel data acquisition mode, collect acceleration signal, and gyroscope signal as sample data. The datasets were preprocessed with deweighting, interpolation, and feature extraction. With the goal of recognizing human posture, experiments with KNN, logistic regression, and random gradient descent algorithms were conducted. In order to verify the superiority of each algorithm, the data window was adjusted to compare the recognition speed, computation time, and accuracy of each classifier. For the problem of improving the accuracy of human posture recognition, a neural network model based on full connectivity is developed. In addition, this paper proposes a finite state machine‐ (FSM‐) based FES control model for controlling the upper limb to perform a range of functional tasks. In the process of constructing the network model, the effects of different hidden layers, activation functions, and optimizers on the recognition rate were experimental for the comparative analysis; the softplus activation function with better recognition performance and the adagrad optimizer are selected. Finally, by comparing the comprehensive recognition accuracy and time efficiency with other classification models, the fully connected neural network is verified in the human posture superiority in identification.