The present paper is motivated by a recent result which is obtained by Liu and Xiao, the exact distribution of roots for a quadratic polynomial will be considered. For completeness, the proof of the Liu-Xiao's lemma will firstly be supplemented, however, the condition of the unit complex roots will be improved. In this case, the computations of the Neimark-Sacker bifurcation or the Hopf bifurcation can be simplified. Furthermore, some new results are also established by using the relationship of the roots and coefficients which is very important in the annals of history, through the discoveries of Abel, Galois and Gauss in the nineteenth century. The obtained results can be used for the bifurcations of the two dimensional discrete-time dynamical systems.
A continuous one-dimension interval map is simple, however, it is very important and that the dynamical behaviors are also very luxuriant, for example, the recurrence sequence $x_{t+1} =ax_{t}(1-x_{t})$ is just a ready-made example. For the identification of the dynamical behaviors for the continuous one-dimension interval map, the Sharkovskii's theorem and the Li-Yorke's theorem are especially important. Thus, in this paper, a review will be given for the convenience of applications.
The true potential of human-AI collaboration lies in exploiting the complementary capabilities of humans and AI to achieve a joint performance superior to that of the individual AI or human, i.e., to achieve complementary team performance (CTP). To realize this complementarity potential, humans need to exercise discretion in following AI 's advice, i.e., appropriately relying on the AI's advice. While previous work has focused on building a mental model of the AI to assess AI recommendations, recent research has shown that the mental model alone cannot explain appropriate reliance. We hypothesize that, in addition to the mental model, human learning is a key mediator of appropriate reliance and, thus, CTP. In this study, we demonstrate the relationship between learning and appropriate reliance in an experiment with 100 participants. This work provides fundamental concepts for analyzing reliance and derives implications for the effective design of human-AI decision-making.
Although pervasive spread of misinformation on social media platforms has become a pressing challenge, existing platform interventions have shown limited success in curbing its dissemination. In this study, we propose a stance-aware graph neural network (stance-aware GNN) that leverages users' stances to proactively predict misinformation spread. As different user stances can form unique echo chambers, we customize four information passing paths in stance-aware GNN, while the trainable attention weights provide explainability by highlighting each structure's importance. Evaluated on a real-world dataset, stance-aware GNN outperforms benchmarks by 32.65% and exceeds advanced GNNs without user stance by over 4.69%. Furthermore, the attention weights indicate that users' opposition stances have a higher impact on their neighbors' behaviors than supportive ones, which function as social correction to halt misinformation propagation. Overall, our study provides an effective predictive model for platforms to combat misinformation, and highlights the impact of user stances in the misinformation propagation.
With the booming development of China's ocean shipping and tourism business, the demand for safety and economy when ships are sailing at sea is increasing. Accurate meteorological and hydrographic forecasts can provide meteorological navigation for ships to avoid typhoons and bad weather areas as much as possible and reduce the damage to the ship's hull from wind and waves. To obtain accurate and reliable wind speed prediction results, this paper combines the advantages of convolutional neural network and gated recurrent unit network to form a deep convolutional gated recurrent unit network model (CNN-GRU). For multiple locations, the CNN-GRU algorithm is used to extract the characteristic meteorological elements, and then the convolutional neural network is used to establish the spatial characteristic relationship between each location, and the gated recurrent unit network is used to establish the temporal characteristic relationship between historical time points, and the final wind speed prediction results are obtained based on the spatio-temporal correlation analysis. In this paper, the CNN-GRU model was established using meteorological data from 2019 to 2021, and the prediction results were compared with CNN model and GRU model and the accuracy was verified. The results show that the experimental results obtained by the CNN-GRU model are more accurate and prove the effectiveness of the proposed method.
Cotton frequently experiences substantial yield reduction due to disease afflictions. This paper introduces a proof-of-concept mobile application that employs transfer learning with pre-trained MobileNetV3 and NASNetMobile models to detect cotton plant diseases. The system addresses a four-class problem, identifying the health status of cotton plants or leaves. Using TensorFlow, these models are refined with publicly available data and deployed on an iOS app. This study also provides a performance assessment of both models, previously unexplored in this specific context. The methodology, showcasing significant potential for real-world implementations, recorded overall accuracies of 97.7% for NASNetMobile and 96.7% for MobileNetV3.
Detecting small objects in aerial images is challenging due to limited information of the object and complex backgrounds surrounding them. Most of the existing object detection methods focus on detecting individual objects and have different feature extraction methods based on object appearance. In this paper, we propose a new Graph Neural Networks (GNN) based method to refine detection results generated by object detectors. In this method, we construct a detection graph by using the predicted detection bounding boxes as nodes, while the features of a bounding box become its node features. Edges are added using distance and topological information. Then, based on the detection confidences of the bounding boxes, some nodes are labeled as bird or non-bird, where nodes with high detection confidences are labelled bird and nodes with very low detection confidences are labeled as non-bird. GNN algorithms are trained using the labeled portion of the graph to infer the classes of the unlabeled nodes. Our experimental results on detecting waterfowl in aerial images show that the new method significantly improved detection accuracy and robustness by significantly reducing false positives.
Internet of Things (IoT) applications suffer from network security, sensor deployment, energy consumption, etc. Recent studies show that most energy is consumed in the case of smart homes applications. Households consume up to 40% of total energy in countries of the European Union. Thus, the improvement of energy management is considered as a huge challenge in IoT application. Machine learning especially deep learning models have been applied to deduce about optimal model which can predict the energy consumption at long term. In this article, we aim to deduce the effective predictive model with less errors in the predictions tests. We propose to compare Gated Recurrent Unit(GRU) and Long Short-Term Memory (LSTM). Therefore, to deduce about the optimal model, several metrics have been evaluated such as R-squared, the minimum mean squared error (MSE), mean absolute error (MAE),root mean squared error (RMSE),etc. These metrics have been applied because they serve different purposes and provide complementary information about the model's accuracy and fit.
This paper presents a frequency offset estimation algorithm based on TRS (tracking reference signal) for 5G NR system in the UE tracking mode. The algorithm makes use of the 4 TRS resources within 2 consecutive slots in 5G NR FR1 to provide a larger frequency tracking range while maintaining a non-decreasing accuracy with a little more calculation complexity.
Time-delay differential equations have important applications in physical chemistry, engineering, information, economy, especially in biological mathematics and many other fields. In this paper, we study the existence of solutions to second-order differential equations with time delays and variation coefficients in a complete b-metric spaces.
In order to objectively evaluate the indicators affecting cultivated land productivity, under the background of big data, taking Jiaozuo City as an example, data mining is used to analyze and analyze the cultivated land productivity indicators, and finally the cultivated land in Jiaozuo City is divided into three types, and four indicators that have an impact on the cultivated land types in Jiaozuo City are screened: average annual rainfall, effective soil thickness, total nitrogen and latitude.
The high-tech electronics industry must comply with international environmental protection and human health regulations, such as the EU RoHS and REACH directives, as well as other regulations that prohibit or restrict certain substances. Electronics and electrical manufacturers have developed green supply chain management systems (GSCMS) to meet these regulatory requirements. However, as the number of GSCMS private cloud systems increases, it becomes more difficult for upstream suppliers to respond to multiple customers' GSCMS investigation processes. The study aims to develop and design a multi-enterprise service platform with a hybrid cloud architecture based on the design science research method (DSRM) to address this issue. The platform's objective is to enhance the provision of cross-system services and both upstream and downstream connection services, to satisfy the material and part approval demands of the industry and its multi-tier suppliers. Moreover, it seeks to minimize the necessity of repeating the green material and part approval process within GSCMS. Developing a green supply chain platform using the DSRM is considered feasible in the study. This platform can effectively lower operating costs for businesses, enhance economic gains, and assist in complying with EU environmental regulations, thereby promoting sustainable development.
Fine-grained Image Classification (FGIC) is a hot research topic in computer vision. Currently, FGIC faces several challenges, such as similar appearances, cluttered backgrounds, and pose variations. To effectively address these challenges, we propose a framework called Scale-Aware Graph Convolutional Network (SAGCN) to capture subtle differences in images. Leveraging the characteristics of fine-grained images, we design two core modules, namely Scale-Aware Selection Module (SASM) and Spatial Semantic Correlation Module (SSCM). SASM aggregates multi-scale information of fine-grained images by fusing features from multiple layers. SSCM establishes semantic-spatial relationships by propagating information among different parts of the fine-grained image. Furthermore, we propose a Pairwise Appearance Similarity Loss (PAS-Loss) to distinguish easily confused categories. Extensive experiments demonstrate that our method achieves state-of-the-art results on benchmark datasets.
In this paper, we describe the process of building a corpus for Tunisian Speech Emotion Recognition (SER). To the best of our knowledge, it is the first work in the SER field that uses spontaneous speech emotion in Tunisian dialect. SER represents an active research problem in the field of Natural Language processing (NLP). It aims to detect different emotions such as satisfaction, frustration and anger from audio speeches using various classifiers. Speech signal preprocessing is the first and the most important step in the SER process. Moreover, Pre-Processing of Speech is very crucial in the applications where silence or ambient noise is completely undesirable. Voice activity detection is a common procedure that plays a key role in preprocessing speech signals and noise cancellation. Pre-emphasis of speech helps the system be computationally more effective [1].This work proposed a preprocessing method to extract features from natural Tunisian speech. Speech preprocessing consists of cleaning the speech signal from ambient and unwanted noises, detecting speech activity and normalizing the length of the vocal tract.
Recognizing the emotional content of Natural Language sentences can improve the way humans communicate with a computer system by enabling them to recognize and imitate emotional expressions. In this paper, deep learning, deep neural networks as well as Transformers were examined. Specifically, we designed and developed deep learning methods and BERT-based implementations for recognizing emotional content in user-generated data. Extensive experiments were conducted using these models on a variety of textual data and all the designed methods were evaluated. The results of the study show that BERT achieved the best overall performance, getting an F1-score of 0.86 on the Twitter Data dataset. Also, the Bi-LSTM with Attention Mechanism and Bi-LSTM using Word2vec performed quite well, achieving F1-scores of 0.83 and 0.82, respectively.
In the age of artificial intelligence (A.I.), software engineering is facing unprecedented changes. Software developers need to have a deep understanding of, especially large model technologies, since the traditional software development model cannot meet the new needs. Moreover, software engineering also needs to pay more attention to the value of data. The data-driven software development models are growing, and data analysis and machine learning technologies have also been widely used. Software development requires higher efficiency, quality, and flexibility. New methods such as agile development and DevOps have emerged. Software testing also needs to be more intelligent, and test automation has become an essential part in software engineering. This speech focuses on sharing the opportunities and challenges brought by GPT and other big models to software development and testing. It also looks forward to the changes brought by A.I. to software engineering education and how we coped. The reform of software engineering is an inevitable trend, and software developers need to constantly learn new technologies and master new methods in the age of A.I.
Take a modified Leslie-Gower predator model as the main body, and the added interference random noise and Gilpin-Ayala schemes. Firstly, the global positive solutions and uniqueness were studied. Then the boundary conditions of species extinction, the persistence and global stability of the mean of the number of species were studied. Meanwhile the numerical simulations verified that the results proved by theoretical studies are consistent with the actual situation.
Numerous models with outstanding performance for detecting typical water surface targets emerged in the first “Ship-Sea Data Intelligence Application Innovation Competition”. By analyzing the differences between these models and their baseline models, a series of improvement strategies were summarized and applied, including improved non-maximum suppression processing, data augmentation methods, and model training processes. Furthermore, experimental analysis results showed a significant increase in the average detection accuracy of the improved models, providing technical support for intelligent and unmanned situation awareness needs on ocean platforms. The team also constructed the first large-scale high-definition ship dataset in China, which includes three types of scenes including dock, lake and ocean-going as well as finely annotated data for various water surface environments such as occlusion, rain, fog, and blur.
