Intelligent Analysis Research Articles

The Application of the Devil’s Advocacy Technique to Intelligence Analysis

Understanding evolving situations and predicting future events are significant challenges in the intelligence sphere. These necessitate advanced techniques for intelligence collection and robust evaluation procedures to enhance the accuracy of predictions. The Devil’s Advocacy technique can be employed to improve analysis procedures by challenging prevailing views and mitigating the risk of cognitive biases. This paper examines the effectiveness of the Devil’s Advocacy technique when applied to intelligence analysis, by evaluating its strengths, weaknesses, and applicability, through a comprehensive literature review and empirical research involving experienced practitioners in New Zealand. Previous studies examining the Devil’s Advocacy technique primarily focused on its contribution to decision-making on the final intelligence product and on the perceptions of the devil’s advocate. These are important, but this paper is also concerned with the analysts’ experience of practically applying the technique for intelligence analysis. This paper contends that the Devil’s Advocacy technique is an attractive complementary technique which is effective in specific situations. It also reveals that New Zealand intelligence practitioners define, perceive, and apply the Devil’s Advocacy technique differently across their organisations. It therefore argues that New Zealand practitioners may have an inconsistent understanding of the technique, indicating potential gaps in the training and education of New Zealand intelligence professionals. However, the inherently subjective nature of the technique itself could also be a contributing factor.

Efficient Pulp and Paper Wastewater Treatment Using Green Tea-Synthesized Magnetite: Artificial Intelligence and Kinetic Analysis

The pulp and paper industry generates pollutant-laden wastewater, requiring effective treatment to prevent environmental damage and enable sustainable reuse. This study investigated the use of green tea-synthesized magnetite (GT-Fe3O4) for treating pulp and paper wastewater. Characterization techniques, including SEM, XRD, FTIR and EDX, were employed to analyze the synthesized magnetite. Batch adsorption studies were conducted to examine the optimal operating conditions, including pH, contact time, adsorbent dose, and stirring rate, for achieving the best treatment performance. The treatment focused on reducing chemical oxygen demand (COD), total suspended solids (TSS), and turbidity in the wastewater. The adsorption data were evaluated using kinetic models, with the pseudo-second-order model providing the best fit. Additionally, an artificial intelligence prediction model (AIP) with a 4–10–1 structure was employed to predict the removal efficiencies. The optimal conditions for the treatment process were determined to be a pH of 6, a contact time of 60min, a GT-Fe3O4 dose of 0.7g/L, and a stirring rate of 150rpm at room temperature. Under these conditions, the removal efficiencies for COD, TSS, and turbidity were 70.5%, 65.4%, and 38.8%, respectively. The AIP model achieved an R² value range of 0.978 to 0.982, showing high prediction accuracy. The study demonstrates that GT-Fe3O4 effectively reduces pollutants in pulp and paper wastewater.

Improved GAN-based deep learning approach for strain field prediction and failure analysis of precast bridge slab joints

Nonlinear finite element analysis (FEA) is crucial for understanding complex stress-strain behaviors and assessing potential structural failure risks. Building on the recent success of deep learning (DL) methods in physical field predictions, there has been growing interest in forecasting internal force fields of structures beyond the existing data scope. In this study, a data-driven strain field analysis model based on an improved Generative Adversarial Network (GAN) is proposed. To enrich the input data, this novel approach incorporates the Signed Distance Field (SDF) derived from the characteristics of the original Finite Element (FE) models. The generator, based on U-Net, is enhanced with channel attention mechanisms and loss penalty terms, forming the core of the improved GAN. The effectiveness of the proposed method is demonstrated using two parallel datasets comprising 3-Headed bar tension models across two planes. Subsequently, the generated strain field is subjected to feature extraction using the 2D-Principal Component Analysis (2D-PCA) method, followed by structural failure determination through the K-Nearest Neighbor (KNN) algorithm. The results indicate that the proposed improved approach achieves a significant reduction in Root Mean Square Error (RMSE) of 20.71 % and 13.21 % across the two datasets, with the structural failure determination method reaching an F1-score of 0.971. Thus, the intelligent strain field analysis method introduced here effectively predicts strain distributions and promptly identifies failure states, offering valuable insights for further numerical analysis and structural design.

Compassionate Principlism: Towards a Novel Alternative to Standard Principlism in Bioethics.

Principlism appears to be the prevailing applied ethical framework in bioethics. Despite the view's various strengths, critics point out that since the principles are ad hoc, conflicts indubitably emerge leading to inconsistency. There is debate around whether principlism can provide definitive action-guiding moral prescriptions or only help structure intelligent analyses and justifications of moral choices. In this paper, I contend that applying concepts of moral symmetry and moral asymmetry allows us to modify one of principlism's principles-the principle of beneficence-into what I will call the principle of compassion. I argue that the principle of compassion can function as an arbitrating or primary principle within the principlist framework. The result is a view we might call compassionate principlism. Arguably, compassionate principlism leads to fewer inconsistencies and provides more acceptable action-guiding moral prescriptions than traditional principlism.

The Future of Cyber Threat Intelligence: Anticipating and Preparing for Evolving Threats

Modern cybersecurity protocols face unprecedented hurdles from the dynamic cyber threat ecology. In this context, the value of cyber threat intelligence as a prophylactic protective strategy increase (Abu, Selamat, Ariffin , & Yusof, 2018). This study provides a futuristic analysis of cyber threat intelligence, emphasizing the vital role that it will play in anticipating and becoming ready for evolving threats. This article also looks at threat actor methods, attack paths, and emerging technologies in an attempt to illustrate the dynamic nature of cyber threats and the need for adaptive intelligence solutions (Abu, Selamat, Ariffin , & Yusof, 2018). Ethical and legal considerations are also explored to highlight the necessity for a thorough and responsible approach to the gathering and implementation of cyber threat intelligence (Wagner, 2019). Through this thorough examination, readers will get a substantial comprehension of the technical developments and strategic requirements that will impact how cyber threat intelligence is developed (Abu, Selamat, Ariffin , & Yusof, 2018) (Wagner, 2019). This will provide enterprises with the agility and resilience to predict, mitigate, and respond to future cyber-attacks.

Improving Landslide Susceptibility Prediction in Uttarakhand through Hyper-Tuned Artificial Intelligence and Global Sensitivity Analysis

Improving Landslide Susceptibility Prediction in Uttarakhand through Hyper-Tuned Artificial Intelligence and Global Sensitivity Analysis

Cultural heritage preservation in the digital age, harnessing artificial intelligence for the future: a bibliometric analysis

PurposeThis study aims to investigate the performance analysis, science mapping and future direction of artificial intelligence (AI) technology, applications, tools and software used to preserve, curate and predict the historical value of cultural heritage.Design/methodology/approachThis study uses the bibliometric research method and utilizes the Scopus database to gather data. The keywords used are “artificial intelligence” and “cultural heritage,” resulting in 718 data sets spanning from 2001 to 2023. The data is restricted to the years 2001−2023, is in English language and encompasses all types of documents, including conference papers, articles, book chapters, lecture notes, reviews and editorials.FindingsThe performance analysis of research on the use of AI to aid in the preservation of cultural heritage has been ongoing since 2001, and research in this area continues to grow. The countries contributing to this research include Italy, China, Greece, Spain and the UK, with Italy being the most prolific in terms of authored works. The research primarily falls under the disciplines of computer science, mathematics, engineering, social sciences and arts and humanities, respectively. Document types mainly consist of articles and proceedings. In the science mapping process, five clusters have been identified. These clusters are labeled according to the contributions of AI tools, software, apps and technology to cultural heritage preservation. The clusters include “conservation assessment,” “exhibition and visualization,” “software solutions,” “virtual exhibition” and “metadata and database.” The future direction of research lies in extended reality, which integrates virtual reality (VR), augmented reality (AR) and mixed reality (MR); virtual restoration and preservation; 3D printing; as well as the utilization of robotics, drones and the Internet of Things (IoT) for mapping, conserving and monitoring historical sites and cultural heritage sites.Practical implicationsThe cultural heritage institution can use this result as a source to develop AI-based strategic planning for curating, preservation, preventing and presenting cultural heritages. Researchers and academicians will get insight and deeper understanding on the research trend and use the interdisciplinary of AI and cultural heritage for expanding collaboration.Social implicationsThis study will help to reveal the trend and evolution of AI and cultural heritage. The finding also will fill the knowledge gap on the research on AI and cultural heritage.Originality/valueSome similar bibliometric studies have been conducted; however, there are still limited studies on contribution of AI to preserve cultural heritage in wider view. The value of this study is the cluster in which AI is used to preserve, curate, present and assess cultural heritages.

Utilization of transformer model in multimodal data fusion learning: Cross-modal knowledge transfer in the new generation learning space

In response to the difficulties in integrating multimodal data and insufficient model generalization ability in traditional cross-modal knowledge transfer, this article used the Transformer model to explore it in the new generation learning space. Firstly, the article analyzed the processing methods of data and models in cross-modal knowledge transfer, and explored the application of Transformer models in the learning space. This model used natural language processing to represent and extract textual features, Mel Frequency Cepstral Coefficients (MFCCs) to represent and extract audio features, and Faster R-CNN (Faster Region-based Convolutional Neural Network) to represent and extract image features. The article also discussed the implementation process of the Transformer model functionality. The experiment used data from four datasets, including Quora Question Pairs, to test the performance of the model’s cross-modal knowledge transfer through intelligent question answering and task analysis. In single type data testing, the accuracy and recall of the model in this article were better than the comparison model in the three types of data. The highest accuracy and recall in the test set were 91% and 93%, respectively. In the most challenging multimodal intelligent question answering test, the speech-image question answering method achieved an accuracy rate of 89% in answering open questions, indicating that the model had good multimodal data fusion ability. In the analysis experiment of 6 homework prone knowledge points on images with text annotations, the induction accuracy reached 85%, indicating that the model had strong generalization ability. The experimental results showed that the Transformer model had good cross-modal knowledge transfer performance, providing a reference for subsequent research on cross-modal knowledge transfer in the new generation learning space.

AI-Led Healthcare Leadership: Unveiling Nursing Trends and Pathways Ahead

Background: Artificial intelligence (AI) is transforming healthcare systems by improving operational efficiency, simplifying patient care procedures, and improving diagnostic accuracy. Artificial intelligence (AI) technologies, like machine learning and natural language processing, present previously unheard-of chances to quickly and accurately evaluate enormous volumes of healthcare data, assisting with clinical decision-making and enhancing patient outcomes. Aim thorough examination and analysis of artificial intelligence's impact on healthcare leadership, with a particular emphasis on present nursing trends and their implications for the future. The study tries to uncover the advantages, difficulties, and consequences of AI integration by looking at how AI technologies including clinical decision support systems, predictive analytics, robots, natural language processing, and telehealth are being used in nursing practice. Method: A comprehensive analysis including research articles published between 2015 and 2024 was carried out. To give a thorough overview of AI's present and future uses in healthcare, major themes and trends were found and summarized. Results: By stressing AI's role in improving diagnostic accuracy and patient outcomes, the study highlights the technology's major contributions to drug discovery, virtual patient care, and medical imaging. Human-centered design concerns, the necessity of educational changes, and ethical challenges surrounding the application of AI surfaced as crucial topics needing attention. Conclusion: AI has enormous potential to transform healthcare by enhancing operational effectiveness, optimizing the delivery of care, and increasing diagnostic precision. Still, ethical issues must be resolved, interdisciplinary cooperation must be promoted, and educational frameworks must be improved in order to provide healthcare workers with the necessary AI skills.

Convolutional neural network as an efficient alternative to supervised learners for modelling plant diseases

The high productivity of the agricultural sector is crucial to the Global economy, as well as the national economy. For instance, India increased its nation's GDP by a factor of 17% to 18% in the past few decades. In India, the farming sector is the primary source of income. Numerous insects and illnesses have an impact on plant development, amount, and quality of products. Therefore, it is crucial to find the diseases early in the plant's development. Image processing is used to identify plant diseases and pests. Artificial intelligence, machine learning, and visual analysis have all been used in the past few decades to find and diagnose plant diseases. These computerized techniques are excellent for quickly monitoring vast acreage. In this study, we used the foliage of potato plants to identify diseased leaves affected by early blight, as well as late blight. A Comparison is thus laid between the Convolutional Neural Network (CNN) Model for this purpose, and the standard Supervised Learning Classifiers, like k – Nearest Neighbors, Support Vector Machine, AdaBoost, etc. The comparison is subjected not only towards the traditional Metrics, like Accuracy, Precision, etc. but is expanded to temporal domains as well, that is the time taken to compute the results. Also, the training, and testing split used while incorporating the model is another deciding factor in its performance. Several such training–testing splits are considered for such, and thereby, the convolutional Neural Network Model is found to be better in performance than the Supervised Machine Learners irrespective of the Splitting Ratio of the Training, and the Testing Data.

A deep learning-powered intelligent microdroplet analysis workflow for in-situ monitoring and evaluation of a dynamic emulsion

Accurately determining microdroplet size and uniformity in dispersed phase systems is key for assessing emulsion stability, and essential for quality monitoring and control of emulsion products. Microscopic imaging probes have provided intuitive and advanced insights for characterizing particle microscopic morphology and dispersion in multiphase flows. However, wide size distributions, edge-cutting, overlapping, and high-density target droplets can compromise the image processing accuracy, thereby affecting the precise assessment of emulsion status. In this study, we present a deep learning-based intelligent workflow for microdroplet characterization using an in-situ high-speed particle imaging probe combined with advanced image processing techniques. With the assistance of progressive automatic annotation, a droplet image database was constructed for training the Mask R-CNN model. By incorporating image patching and supervised data augmentation strategies, our well-trained model with accuracy of 95.5% can precisely detect, localize, and segment microscale droplets with various patterns. Additionally, the shape reconstruction module visualizes true shapes of overlapping and edge-cutting droplets, facilitating precise quantification of droplet size information. To validate the feasibility and generalizability of the proposed workflow, we obtained sufficient droplet images through offline sampling during emulsion processes while employing back and front illumination for in-situ monitoring. The automated method achieves precise droplet detection and size measurement, demonstrating its immense potential in online monitoring, evaluation, and control of emulsion quality.

Accelerated intelligent prediction and analysis of mechanical properties of magnesium alloys based on scaled super learner machine-learning algorithms

The use of machine learning algorithms in magnesium (Mg) alloys has evolved a scientific innovation for lightweight. The dataset was compiled by collecting data from the experiment and utilizing machine learning (ML) models to predict the mechanical properties of 348 Mg alloys. The proportion between the predicted and experimental results produced by different ML models demands more advanced regression methods to obtain better results. Utilizing Mg alloy descriptors as input variables and mechanical properties as output variables, four different ML models were employed namely (i.e.) Random Forest (RF), Extra Tree (ET), Gradient Boost (GB), and Extreme Gradient Boost (XGBoost) to resolve this difficult problem. Each single algorithm aimed to predict the mechanical properties of Mg alloy i.e. Ultimate Tensile Strength (UTS), Yield Strength (YS), and Elongation (EL). Subsequently, the data-driven intelligent prediction modeling technique called scaled Super Learner (SL) was employed to integrate the single models into the stacked model approach to enhance prediction accuracy. The results obtained using scaled Super Learner demonstrated enhanced prediction accuracy for UTS, YS, and EL. The findings further demonstrate enhanced prediction ability by outperforming other approaches as demonstrated by lower Root Mean Squared Error (RMSE) and higher R-Squared (R2) compared to previous studies. The reason for choosing Scaled Super Learner is because of its robustness and resistance to overfitting. Scaled Super Learner is also widely known for its better scalability, simplicity, and ability to handle noisy. The scaled Super Learner is an optimal approach for predicting the properties of Mg alloys. The proposed scaled Super learner serves as a tool for predicting Mg alloy properties.

TTPXHunter: Actionable Threat Intelligence Extraction as TTPs from Finished Cyber Threat Reports

Understanding the modus operandi of adversaries aids organizations to employ efficient defensive strategies and share intelligence in the community. This knowledge is often present in unstructured natural language text within threat analysis reports. A translation tool is needed to interpret the modus operandi explained in the sentences of the threat report and convert it into a structured format. This research introduces a methodology named TTPXHunter for automated extraction of threat intelligence in terms of Tactics, Techniques, and Procedures (TTPs) from finished cyber threat reports. It leverages cyber domain-specific state-of-the-art natural language model to augment sentences for minority class TTPs and refine pinpointing the TTPs in threat analysis reports significantly. We create two datasets: an augmented sentence-TTP dataset of \(39,296\) sentence samples and a \(149\) real-world cyber threat intelligence report-to-TTP dataset. Further, we evaluate TTPXHunter on the augmented sentence and report datasets. The TTPXHunter achieves the highest performance of \(92.42\%\) f1-score on the augmented dataset, and it also outperforms existing state-of-the-art TTP extraction method by achieving an f1-score of \(97.09\%\) when evaluated over the report dataset. TTPXHunter significantly improves cybersecurity threat intelligence by offering quick, actionable insights into attacker behaviors. This advancement automates threat intelligence analysis and provides a crucial tool for cybersecurity professionals to combat cyber threats.

Smartphone-Assisted Nanozyme Colorimetric Sensor Array Combined "Image Segmentation-Feature Extraction" Deep Learning for Detecting Unsaturated Fatty Acids.

Conventional methods for detecting unsaturated fatty acids (UFAs) pose challenges for rapid analyses due to the need for complex pretreatment and expensive instruments. Here, we developed an intelligent platform for facile and low-cost analysis of UFAs by combining a smartphone-assisted colorimetric sensor array (CSA) based on MnO2 nanozymes with "image segmentation-feature extraction" deep learning (ISFE-DL). Density functional theory predictions were validated by doping experiments using Ag, Pd, and Pt, which enhanced the catalytic activity of the MnO2 nanozymes. A CSA mimicking mammalian olfactory system was constructed with the principle that UFAs competitively inhibit the oxidization of the enzyme substrate, resulting in color changes in the nanozyme-ABTS substrate system. Through linear discriminant analysis coupled with the smartphone App "Quick Viewer" that utilizes multihole parallel acquisition technology, oleic acid (OA), linoleic acid (LA), α-linolenic acid (ALA), and their mixtures were clearly discriminated; various edible vegetable oils, different camellia oils (CAO), and adulterated CAOs were also successfully distinguished. Furthermore, the ISFE-DL method was combined in multicomponent quantitative analysis. The sensing elements of the CSA (3 × 4) were individually segmented for single-hole feature extraction containing information from 38,868 images of three UFAs, thereby allowing for the extraction of more features and augmenting sample size. After training with the MobileNetV3 small model, the determination coefficients of OA, LA, and ALA were 0.9969, 0.9668, and 0.7393, respectively. The model was embedded in the smartphone App "Intelligent Analysis Master" for one-click quantification. We provide an innovative approach for intelligent and efficient qualitative and quantitative analysis of UFAs and other compounds with similar characteristics.

DeepInvesting: Stock market predictions with a sequence-oriented BiLSTM stacked model – A dataset case study of AMZN

Intelligent forecasters are now being considered in the stock market, providing essential insights and strategic guidance to investors and traders by presenting analytical tools and predictive models, thus enabling informed decision-making and mitigating financial risks in this dynamic market. The importance of intelligent analyzers in stock trading routines is considered in this work, where DeepInvesting, a multimodal deep learning model tailored for stock price prediction, is introduced. Employing a Sequence-Oriented, Long-Term Dependent (SoLTD) architecture featuring Bidirectional Long Short-Term Memory (BiLSTM) networks, DeepInvesting is applied to essential features of the Amazon Corp. (AMZN) market dataset, gathered from Yahoo Finance, including Closing, Opening, High, Low, Volume, and Adj Close prices. Exceptional performance in forecasting Closing, Opening, High, Low, and Adj Close prices is demonstrated, with minimal Mean Absolute Percentage Error (MAPE) and Root Mean Squared Percentage Error (RMSPE) scores, coupled with high R-squared (R2) values, manifesting a robust fit to the data, as well as computational complexity, and Rates Per Second (RPS) metrics in comparison to other models of KNN, LSTM, RNN, CNN, and ANN. Finally, challenges in the accurate prediction of trading volumes are identified, highlighting an area for future enhancement.

MAKG: A maritime accident knowledge graph for intelligent accident analysis and management

With the increasing frequency of human activities at sea, maritime accidents are occurring more often. Analyzing and mining maritime accident cases can help uncover the causal mechanisms behind these incidents, thereby enhancing maritime safety. As an emerging technology for knowledge management and mining, knowledge graphs offer significant support for the storage, reasoning, and decision-making processes related to maritime accidents.In this study, we established a knowledge graph construction and application framework for maritime accidents to facilitates the extraction and management of maritime knowledge from unstructured texts. First, 581 accident reports released by the China Maritime Safety Administration over the past decade (2014–2023) were used as the data basis for analysis and construction of the maritime accident ontology structure using the seven-step method, which comprises 8 entity types, 8 relationship types, and 18 attribute entity types. Second, We proposed MBERT-BiLSTM-CRF-SF, a named entity recognition model based on domain pretraining and self-training, to reduce graph construction costs. This model achieved state-of-the-art performance in the maritime domain, with an F1 score of 0.910 ± 0.006, which is about 5% higher than the mainstream model. In addition, we proposed an entity alignment method based on font and semantics to refine knowledge further. On the basis of the proposed method, we constructed a large, high-quality maritime accident knowledge graph (MAKG) system that contains 16,099 entities and 20,809 relationship instances. Finally, we reduced the complexity of applying knowledge graphs by integrating the CRISPE prompt learning framework of the large language model, and experiments on graph traversal, pattern recognition, and aggregation analysis were conducted to assess the quality of MAKG. Results demonstrate that MAKG can effectively enhance the efficiency of querying and reasoning about maritime accident information, thus providing significant support for the prevention and management of maritime accidents.

Intelligent Analysis of Avian Influenza Virus Biomarkers Based on Chemiluminescence Resonance Energy Transfer and DNA Logic Analysis.

A one-step logical analysis of multiple targets remains challenging. Herein, we report a one-pot and intelligent DNA logical analysis platform for the diagnosis of avian influenza virus (AIV) biomarkers based on chemiluminescence resonance energy transfer (CRET) and logic operations. On the surface of Lum/PEI/CaCO3 microparticles, the excited state of luminol underwent CRET with fluorescein isothiocyanate or rhodamine B isothiocyanate, producing three well-separated light emissions at 425, 530, and 590 nm, respectively. Taking advantage of the close distance between fluorophores aligned by the catalytic hairpin assembly reaction, the CRET efficiency was greatly enhanced (53.1%). H1N1, H7N9, and H5N1 were detected with limits of detection values as low as 15, 34, and 58 pM, respectively. Three-input logic circuits were simultaneously conducted on the surface of Lum/PEI/CaCO3 microparticles, enabling the rapid and accurate discrimination of multiple AIV biomarkers in one solution. In terms of peak positions and the normalized value of the total peak intensity, three biomarkers can be simultaneously discriminated without any other complex operations. In summary, the CRET-based multiple analytical assay was developed as an intelligent biosensor for identifying AIV biomarkers, having promising application prospects in the field of multiple analysis and precise disease diagnosis.

A Cross Spatio-Temporal Pathology-based Lung Nodule Dataset

Recently, intelligent analysis of lung nodules with the assistant of computer aided diagnosis (CAD) techniques can improve the accuracy rate of lung cancer diagnosis. However, existing CAD systems and pulmonary datasets mainly focus on Computed Tomography (CT) images from one single period, while ignoring the cross spatio-temporal features associated with the progression of nodules contained in imaging data from various captured periods of lung cancer. If the evolution patterns of nodules across various periods in the patients’ CT sequences can be explored, it will play a crucial role in guiding the precise screening identification of lung cancer. Therefore, a cross spatio-temporal lung nodule dataset with pathological information for nodule identification and diagnosis is constructed, which contains 317 CT sequences and 2,295 annotated nodules from 109 patients. This comprehensive database is intended to drive research in the field of CAD towards more practical and robust methods, and also contribute to the further exploration of precision medicine related field. To ensure patient confidentiality, we have removed sensitive information from the dataset.

Analyzing K-12 AI education: A large language model study of classroom instruction on learning theories, pedagogy, tools, and AI literacy

There is growing recognition among researchers and stakeholders about the significant impact of artificial intelligence (AI) technology on classroom instruction. As a crucial element in developing AI literacy, AI education in K-12 schools is increasingly gaining attention. However, most existing research on K-12 AI education relies on experiential methodologies and suffers from a lack of quantitative analysis based on extensive classroom data, hindering a comprehensive depiction of AI education's current state at these educational levels. To address this gap, this article employs the advanced semantic understanding capabilities of large language models (LLMs) to create an intelligent analysis framework that identifies learning theories, pedagogical approaches, learning tools, and levels of AI literacy in AI classroom instruction. Compared with the results of manual analysis, analysis based on LLMs can achieve more than 90% consistency. Our findings, based on the analysis of 98 classroom instruction videos in central Chinese cities, reveal that current AI classroom instruction insufficiently foster AI literacy, with only 35.71% addressing higher-level skills such as evaluating and creating AI. AI ethics are even less commonly addressed, featured in just 5.1% of classroom instruction. We classified AI classroom instruction into three categories: conceptual (50%), heuristic (18.37%), and experimental (31.63%). Correlation analysis suggests a significant relationship between the adoption of pedagogical approaches and the development of advanced AI literacy. Specifically, integrating Project-based/Problem-based learning (PBL) with Collaborative learning appears effective in cultivating the capacity to evaluate and create AI.

A Framework for Agricultural Intelligent Analysis Based on a Visual Language Large Model

Smart agriculture has become an inevitable trend in the development of modern agriculture, especially promoted by the continuous progress of large language models like chat generative pre-trained transformer (ChatGPT) and general language model (ChatGLM). Although these large models perform well in general knowledge learning, they still have certain limitations and errors when facing agricultural professional knowledge about crop disease identification, growth stage judgment, and so on. Agricultural data involves images and texts and other modalities, which play an important role in agricultural production and management. In order to better learn the characteristics of different modal data in agriculture, realize cross-modal data fusion, and thus understand complex application scenarios, we propose a framework AgriVLM that uses a large amount of agricultural data to fine-tune the visual language model to analyze agricultural data. It can fuse multimodal data and provide more comprehensive agricultural decision support. Specifically, it utilizes Q-former as a bridge between an image encoder and a language model to achieve a cross-modal fusion of agricultural images and text data. Then, we apply a Low-Rank adaptive to fine-tune the language model to achieve an alignment between agricultural image features and a pre-trained language model. The experimental results prove that AgriVLM demonstrates great performance in crop disease recognition and growth stage recognition, with recognition accuracy exceeding 90%, demonstrating its capability to analyze different modalities of agricultural data.