Inteligencia Artificial

TVN: Detect Deepfakes Images using Texture Variation Network

Haseena S — 2023-05-24

Face manipulation technology is rapidly developing, making it impossible for human eyes to recognize fake face photos. Convolutional Neural Network (CNN) discriminators, on the other hand, can fast achieve high accuracy in distinguishing fake/real face photos. In this paper, we look at how CNN models discern between fake and real faces. Face forgery detection relies heavily on Texture Variation Network (TVN) information, according to our findings. We propose a new model, TVN, for robust face fraud detection, based on Convolution and pyramid pooling (PP), as a result of the aforesaid discovery. To produce a stationary representation of composition difference information, Convolution combines pixel intensity and pixel gradient information. Simultaneously, multi-scale information fusion based on the PP can prevent the texture features from being destroyed. Our TVN beats previous techniques on numerous databases, including Faceforensics++, DeeperForensics-1.0, Celeb-DF, and DFDC. The TVN is more resistant to image distortion, such as JPEG compression and blur, which is critical in the wild.

PB3C-CNN: An integrated PB3C and CNN based approach for plant leaf classification

Sukanta Ghosh — 2023-05-24

Plant identification and classification are critical to understand, protect, and conserve biodiversity. Traditional plant classification requires years of intensive training and experience, making it difficult for others to classify plants. Plant leaf classification is a challenging issue as similar features appears in different species of plant. With the development of automated image-based classification, machine learning (ML) is becoming very popular. Deep learning (DL) methods have significantly improved plant image identification and classification. In the last decade, convolutional neural networks (CNN) have entirely dominated the field of computer vision, showing outstanding feature extraction capabilities and significant identification and classification performance. The capability of CNN lies in its network. The primary strategy to continue this trend in the literature relies on further scaling networks in size. However, costs increase rapidly, while performance improvements may be marginal when the number of net-works increases. Hence, there is a need to optimize the CNN network to get the best possible result with the minimum number of networks and other parameters such as the number of epochs, number of layers, batch size and number of neurons. The paper aims to evolve the optimal architecture of CNN using PB3C algorithm for plant leaf classification. For this, we use the nature-inspired computing technique parallel big bang–big crunch to evolve a CNN's optimal architecture automatically. Current study validated the proposed approach for plant leaf classification and compared it with 11 other machine learning-based approaches. From the results obtained it was found that the proposed approach was able to outperforms all 11 existing state-of-the-art techniques.

New Exploration/Exploitation Improvements of GWO for Robust Control of a Nonlinear Inverted Pendulum

Soumia Mohammed Djaouti — 2023-06-10

Tuning a nonlinear inverted pendulum is a complex and uncertain optimization problem. In this paper, we develop two new GWO variants by introducing a DLH (Dimension Learning-based Hunting) module and new formulas to enhance the exploitation/exploration ratio aiming to avoid local minima. A statistical analysis is carried out to compare the two proposed approaches with five GWO variants. After that, they are used to tune a PID and FSMC controller. The obtained results are promising even when compared to other approaches

An Optimized Clustering Approach using Tree Seed Algorithm for the Brain MRI Images Segmentation

ghazi boumediene ghaouti — 2023-06-10

Clustering algorithms are widely used to segment medical images. However, these techniques are difficult to perform, especially in brain magnetic resonance images (MRI), given the complexity of the anatomical structure of brain tissue, the in-homogeneity of pixel intensity in these images, and partial volume and noise effects. This will cause the algorithm to fall into the local minima problem; for this reason, it is recommended to improve such clustering algorithms using optimization techniques to obtain better results. In this study, we have proposed a developed clustering algorithm and we optimized it using a tree seed algorithm (TSA) to segment brain MRI image. Algorithms are tested on real brain image datasets. The experimental results on simulated and real brain MRI datasets show that our proposed method has satisfactory results regarding the Davies-Bouldin index (DBI) compared to the fuzzy c-mean (FCM) algorithm.

Exploring the Generality of Norms in Multi-Agent Systems

Jhonatan Alves — 2023-06-16

Norms are useful tools to regulate autonomous agents, and their generality is the focus of this paper. The generality of norms refers to the extent of behaviors the norms are capable of regulating. While very specific norms tend to be inefficient to avoid undesirable behaviors (since they are rarely activated), very general norms tend to limit excessively the options of the agents (since they are activated too often) hindering them to achieve the system goal. Therefore, a norm that efficiently regulates the agents should have a balanced generality, being neither too specific nor too general. Therefore, we consider that exploring the generality of norms is a fundamental key to obtaining efficient norms. However, the evaluation of their generality usually considers every behavior they regulate. Since it is likely an unfeasible task, in this paper, we investigate alternatives to estimate the norms generality from their syntactic characteristics. Based on these characteristics, we obtain different sequences of norms that vary, approximately, from the most specific to the most general. We assume thus that norms with a balanced generality are more easily found considering these orderings. Therefore, it is relevant to understand the impact of the syntactical characteristics in ordering the norms. In this context, we found out how different alternatives organize the norms space. This result is particularly useful for the development of algorithms for searching efficient norms that, through different strategies, may exploit how norms space is arranged and may be pruned.

Semi-supervised learning models for document classification: A systematic review and meta-analysis

Alex Cevallos-Culqui — 2023-06-09

The continuous increase of digital documents on the web creates the need to search for information patterns that allow the categorization of organizational documents to generate knowledge in an institution. An Artificial Intelligence technique for this purpose is text classification, it for its application uses labels (previously categorized documents) with supervised (with labels) or unsupervised (without labels) training models. Both traditional models with their advantages and disadvantages have been joined into semi-supervised models that extract the best qualities of each one, however, the labeling process involves resources and time that try to be optimized to improve classification accuracy.

An analysis of the different semi-supervised models would show us the advantages of their training and the way how the structure of each of them affects the accuracy of their classification. In the present study, a classification structure of the semi-supervised models in the classification of documents is proposed to analyze their qualities and categorization process, through an SLR (Revision of systematic literature) that extracts performance metrics from the identified studies to perform a meta-analysis through forest plots.

To define the search strategy for studies, the PICOC (Population, Intervention, Comparison, Outputs, Context) method has been used, it is supported by the research question defines a search string, which has allowed the collection of 228 research, these are filtered with the PRISMA declaration method and the determination of exclusion criteria, in this way 35 researches are selected for the present study.

The analysis of the selected studies identifies a structure for the different semi-supervised learning models, and a scheme of their work process is obtained, it has been used to extract advantages, disadvantages, and performance metrics. Through a meta-analysis with forest diagrams, the classification accuracy performance of the researches in each learning model is evaluated, determining as results that regardless of the characteristics of its process, active learning (0.89) and assembled learning (0.83) present the best performance levels.

An Improved BAT Algorithm Using Density-Based Clustering

Samraa Al-Asadi — 2023-08-09

BAT algorithm is a nature-inspired metaheuristic algorithm that depends on the principle of the echolocation behavior of bats. However, the algorithm suffers from being stuck in the local optima early due to its poor exploration. An improved BAT algorithm based on the density-based clustering technique is proposed to enhance the algorithm’s performance.

In this paper, the initial population is improved by generating two populations, randomly and depending on the clusters’ center information, and by getting the fittest individuals from these two populations, the initial improved one is generated. The random walk function is improved using chaotic maps instead of the fixed-size movement, and so the local search is improved as well as the global search abilities by diversifying the solutions. Another improvement is to deal with stagnation by partitioning the search space into two parts depending on the generated clusters’ information to obtain the newly generated solution and comparing their quality with the previously generated solution and choosing the best.

The performance of the proposed improved BAT algorithm is evaluated by comparing it with the original BAT algorithm over ten benchmark optimization test functions. Depending on the results, the improved BAT outperforms the original BAT by obtaining the optimal global solutions for most of the benchmark test functions.

Online Incremental Learning Based on Crowdsourcing For Indonesian Ontology Relation Extraction

Eunike Andriani Kardinata — 2023-09-06

Ontology is one form of structured representation of knowledge. Ontology is widely used and developed in information retrieval because of its ability to represent knowledge in a form that machines and humans can understand. With the increasing scale and complexity of ontology, there are more significant challenges in identifying extra-logical errors. Ontological development methods mostly use machine learning, which is at risk of missed extra-logical errors. To handle it, crowdsourcing is used, i.e. dividing a large job into several small jobs and hiring the masses to complete it. Data processing is usually done offline to take advantage of crowdsourcing, and batches are converted into online and incremental. Online incremental learning directly arranges an iterative model after a change is made by ensuring that the knowledge that has been obtained before is maintained. This study built an interactive medium to present the initial relationship between concept pairs. Crowdsourcing participants were asked to validate the relationship repeatedly until a specified accuracy value was reached. This study found that the crowdsourcing process was able to improve the model used in the relationship extraction process, from F1-Score 87.2% to 89.8%. Improvements using crowdsourcing achieve the same result as improvements by experts. Thus, crowdsourcing can correct extra-logical errors appropriately as an expert. In addition, it was also found that offline incremental learning using Random Forest resulted in higher model accuracy than incremental online learning using Mondrian Forest. The accuracy of the Random Forest model has a final accuracy of 90.6%, while the accuracy of the Mondrian Forest model is 89.7%. From these results, it was concluded that incremental online learning cannot provide better results than offline incremental learning to improve the meronymy relationship extraction process.

Detection of Post COVID-Pneumonia Using Histogram Equalization, CLAHE Deep Learning Techniques

Vinodhini M — 2023-09-06

Pneumonia, also known as bronchitis, is caused by bacteria, viruses, or fungi. Pneumonia can be fatal to an infected person because the lungs cannot exchange air. The disease primarily affects infants and people over the age of 65. Every year, nearly 4 million people are killed by the disease, which affects an estimated 420 million people. It is critical to detect and diagnose this condition as soon as possible. Diagnosing the condition using the patient's x-ray is the most effective method. Experienced radiologists will use a chest x-ray of the affected patient to make this informed decision. Recently, coronavirus is a contagious viral disease caused by the SARSCoV2 virus. This virus affects the human respiratory system. The virus also causes pneumonia (COVID pneumonia), which is far more dangerous than normal pneumonia. The main purpose of this task is to study and compare several deep learning enhancement techniques applied to medical x-ray and CT scan images for the detection of COVID19 (pneumonia).

A convolutional neural network (CNN) is used to design a model that can distinguish between COVID19 pneumonia and normal pneumonia. In addition, image enhancement techniques (histogram equalization (HE), contrast-limited adaptive histogram equalization (CLAHE)) have been processed against the dataset to find more efficient methods and models for detecting pneumonia. A dataset of 6432 CXRs were used - 576 COVID pneumonia CXRs, 1583 normal pneumonia CXRs, and 4273 healthy lung CXRs. Based on the results, it was observed that the equalized histogram and the equalized dataset of CLAHE run faster than the original dataset. This requires a computer-aided diagnosis (CAD) system that can distinguish between COVID pneumonia, normal pneumonia, and healthy lungs. In addition, the improved VGG16 achieved 96% accuracy in the detection of X-ray images of COVID19 - pneumonia.

Machine Learning Algorithm for a Link Adaptation strategy in a Vehicular Ad hoc Network

Etienne Feukeu — 2023-09-10

Vehicular Ad Hoc Networks (VANETs) were created more than eighteen years ago with the aim of reducing accidents on public roads and saving lives. Achieving this goal depends on VANET mobiles exchanging Road State Information (RSI) with their surroundings and acting on the received RSI. Therefore, it is essential to ensure that transmitted messages are accurately received. This requires controlling the quality of the sharing medium or link while considering Channel State Information (CSI), which provides information on channel quality and Signal to Noise Ratio (SNR). The process of adjusting the payload based on the CSI is known as Link Adaptation (LA). While several LA works have been published in VANETs, few have considered the effect of relative node mobility. This work presents a link adaptation strategy for VANETs that uses a Neural Network (NN) and the Levenberg-Marquardt Algorithm (LMA). While accounting for Doppler Shift effect induced by the relative velocity, the simulation results demonstrate that the NN approach outperforms its peers by 77%, 115% and 853% in terms of the transmitted errors, model efficiency and throughput respectively, compared to Cte, ARF, and AMC algorithms.

An Ensemble Classification Method Based on Deep Neural Networks for Breast Cancer Diagnosis

Yan Gao — 2023-09-14

Advances in technology have led to advances in breast cancer screening by detecting symptoms that doctors have overlooked. In this paper, an automatic detection system for breast cancer cases based on Internet of Things (IoT) is proposed. First, using IoT technology, direct medical images are sent to the data repository after the suspicious person's visit through medical equipment equipped with IoT. Then, in order to help radiologists, interpret medical images as best as possible, we use four pre-trained convolutional neural network models including InceptionResNetV2, InceptionV3, VGG19 and ResNet152. These models are combined by an ensemble classifier. Also, these models are used to accurately predict cases with breast cancer, healthy people, and cases with pneumonia by using two datasets of X-RAY and CT-scan in a three-class classification. Finally, the best result obtained for CT-scan images belongs to InceptionResNetV2 architecture with 99.36% accuracy and for X-RAY images belongs to InceptionV3 architecture with 96.94% accuracy. The results show that this method leads to a reduction in daily visits to medical centers and thus reduces the pressure on the medical care system. It also helps radiologists and medical staff to detect breast cancer in its early stages.

Fake News Detection in Low Resource Languages using SetFit Framework

Amin Abdedaiem — 2023-09-20

Social media has become an integral part of people’s lives, resulting in a constant flow of information. However, a concerning trend has emerged with the rapid spread of fake news, attributed to the lack of verification mechanisms. Fake news has far-reaching consequences, influencing public opinion, disrupting democracy, fueling
social tensions, and impacting various domains such as health, environment, and the economy. In order to identify fake news with data sparsity, especially with low resources languages such as Arabic and its dialects, we propose a few-shot learning fake news detection model based on sentence transformer fine-tuning, utilizing no crafted prompts and language model with few parameters. The experimental results prove that the proposed method can achieve higher performances with fewer news samples. This approach provided 71% F1 score on the Algerian dialect fake news dataset and 70% F1 score on the Modern Standard Arabic (MSA) version of the same dataset, which proves that the approach can work on the standard Arabic and its dialects. Therefore, the proposed model can identify fake news in several domains concerning the Algerian community such as politics, COVID-19, tourism, e-commerce, sport, accidents, and car prices.

An intelligent approach for anomaly detection in credit card data using bat optimization algorithm

Haseena Sikkandar — 2023-09-27

As technology advances, many people are utilising credit cards to purchase their necessities, and the number of credit card scams is increasing tremendously. However, illegal card transactions have been on the rise, costing financial institutions millions of dollars each year. The development of efficient fraud detection techniques is critical in reducing these deficits, but it is difficult due to the extremely unbalanced nature of most credit card datasets. As compared to conventional fraud detection methods, the proposed method will help in automatically detecting the fraud, identifying hidden correlations in data and reduced time for verification process. This is achieved by selecting relevant and unique features by using Bat Optimization Algorithm (BOA). Next, balancing is performed in the highly imbalanced credit card fraud dataset using Synthetic Minority over-sampling technique (SMOTE). Then finally the CNN model for anomaly detection in credit card data is built using full center loss function to improve fraud detection performance and stability. The proposed model is tested with Kaggle dataset and yields around 99% accuracy.

An Automatic Non-Destructive External and Internal Quality Evaluation of Mango Fruits based on Color and X-ray Imaging using Computer Vision Techniques

Vani Ashok — 2023-09-29

Quality evaluation of food products, agricultural produce to be specific, has gained momentum from past few decades due to the increased awareness among consumers across the world. This has resulted in the increased emphasis on the development and use of quality assessment techniques in food industry. Moreover, there is a need to automate the quality monitoring of agricultural produce like fruits and vegetables which is otherwise done manually in developing countries hence labor intensive, time consuming and subjective in nature. This paper presents an empirical analysis to build a rapid, robust, real-time, non-destructive computer vision based quality assessment model for mango fruits. The work employs the automatic disease classification of mango fruits based on machine and deep learning models. Firstly, the dataset of colored mango fruits images with 2279 images falling into three classes and another dataset of soft X-ray images of mango fruits with 572 images belonging to two quality classes are developed for detecting external and internal defects, respectively. The multilayer perceptron neural network (MLP NN) with two hidden layers, which may be considered as the starting point for deep learning technique, is proposed as machine learning model to classify the color images of mango fruits into one of three external quality classes with 95.1% accuracy and also to classify the soft X-ray images into two internal quality classes with 97.5% accuracy. In order to step out of feature engineering, actual deep learning convolutional neural network (CNN) models, a customized CNN model and pre-trained CNN models, VGGNet (VGG16) and DenseNet121 were also explored for mango disease classification. The maximum validation accuracy of custom CNN was found to be with 91.52% and 98.7% for color and augmented X-ray images, respectively. The classification accuracy of pre-trained models were found to be reasonably good for the color images but exhibited high variability in results and made it difficult to draw a general conclusion for the proposed datasets. However, the proposed MLP NN model based on few basic intensity and geometric features and also the proposed customized CNN model were found to be the best models and they outperform the state of the art reported in the literature.

Rumex Weed Classification Using Region-Convolution Neural Networks Based-Colour Space Information

Saleh Nazal — 2023-11-02

Weed detection is considered the gold standard in smart agriculture field. An automated detection of weed
procedure is a complicated task, specifically detection of Rumex weed due to different real-world environmental conditions, including illumination, occlusion, overlapped, growth stage, and colours. Few works have done
to classify Rumex weed using machine learning. However, the performance is still not at the level required for
agriculture communities and challenges have not been solved. This work proposes Region-Convolutional Neural
Networks (RCNNs) and VGG16 model based on colour space information to classify Rumex weed from grassland.
This paper is investigated the effectiveness of our proposed method over real-world images under different conditions. The findings have shown that the proposed method superior comparing with other AI existing techniques.
The results demonstrate that the proposed method has an excellent adaptability over real-world images.