Traditional methods of responding to cybersecurity incidents often use set playbooks or routines that don’t change, which might not work well with the constantly changing nature of modern cyber dangers. The idea behi...
详细信息
ISBN:
(纸本)9789819601462
Traditional methods of responding to cybersecurity incidents often use set playbooks or routines that don’t change, which might not work well with the constantly changing nature of modern cyber dangers. The idea behind this paper is to create and use decision systems that are specifically designed for adapting how to handle hacking incidents. Advanced artificial intelligence (AI) methods, such as machine learning and natural language processing, are used by these decision systems to analyze security events in real time and help people make quick, well-informed decisions. These systems improve the speed and efficiency of hacking reaction by constantly learning from past events and responding to new threats. Automatic danger recognition, risk assessment, and reaction planning are important parts of the suggested decision systems. Automated systems that look for threats use AI algorithms to look for signs of bad behavior in network traffic, system logs, and other relevant data sources. The risk assessment tools look at the dangers and how bad they could be, considering things like how sensitive the data is, how important the system is, and the rules that need to be followed. Decision systems actively plan responses based on danger recognition and risk assessment results. These responses may include separating affected systems, blocking suspicious IP addresses, or sending events to human experts for further study. Over time, these systems get better at making decisions by using adaptive learning methods and constant feedback loops. This makes it easier for them to spot and stop new cyber threats. The creating and using decision systems for adaptive cybersecurity incident reaction is a big step toward making organizations’ defenses more resilient and effective in the face of changing cyber dangers. By using AI and machine learning, these systems stop threats before they happen, giving cybersecurity teams the tools they need to stay ahead of attackers in the ongoing battl
Multi-modal entity alignment (MMEA) aims to identify equivalent entities across different multi-modal knowledge graphs. In these graphs, entities are enriched with information from various modalities, such as text, im...
详细信息
ISBN:
(数字)9798331508821
ISBN:
(纸本)9798331508838
Multi-modal entity alignment (MMEA) aims to identify equivalent entities across different multi-modal knowledge graphs. In these graphs, entities are enriched with information from various modalities, such as text, images, and numerical data, making the alignment task both challenging and crucial for improving knowledge graph quality. Current MMEA algorithms typically encode entity information separately for each modality using corresponding encoders, and then integrate these representations through various modal fusion strategies. However, these methods often fail to fully exploit the multi-modal information of entities. To address this issue, we propose a feature-enhanced multi-modal entity alignment transformer (FEMEAT). FEMEAT enhances entity attribute information by incorporating modal distribution data, which captures the inherent distribution of different modalities for each entity. This inclusion allows the model to have a richer understanding of entity characteristics across modalities. Additionally, FEMEAT utilizes an Optical Character Recognition (OCR) model to extract and incorporate textual information from images. By integrating this text extracted from images, the model can better utilize the visual modality, enhancing its ability to understand and process multi-modal information. Furthermore, FEMEAT employs a multi-head cross-modal attention (MHCA) mechanism for modal fusion to achieve comprehensive multi-modal entity representation. This mechanism enables the model to attend to different modalities simultaneously and learn a detailed representation of entities by considering the interactions between modalities. The multi-head cross-modal attention mechanism facilitates a nuanced understanding and integration of multi-modal data. Experimental results demonstrate that our model achieves state-of-the-art (SOTA) performance across various training scenarios. The code and datasets used in this study can be accessed at https://***/zewenD/FEMEAT.
The proper detection of brain tumors as quickly as possible and the administration of efficient therapy are both necessary steps in the process of curing patients. The complexities of tumor morphology, such as size, l...
详细信息
ISBN:
(数字)9798350389449
ISBN:
(纸本)9798350389456
The proper detection of brain tumors as quickly as possible and the administration of efficient therapy are both necessary steps in the process of curing patients. The complexities of tumor morphology, such as size, location, and texture, as well as heteromorphic appearance in medical pictures, make tumor analysis a difficult *** three most frequent types of brain cancer are brain tumors, meningiomas, and pituitary gland tumors. The purpose of this study is to investigate a classification issue known as a triple-class problem in order to correctly identify these *** a pre-trained version of the vgg16 network, the proposed classification method pulls characteristics from brain MRI scans. Classifier models that have shown their worth in practice are given to help put the retrieved attributes into meaningful groups. The presented method is carried out on the Python platform, and its efficacy is evaluated with the use of performance *** to the results, the conventional neural network (CNN) technique was outperformed by the Transfer Learning method, which resulted in the model having a higher overall performance. This can be observed by demonstrating that the model had a stronger overall performance.
The precise prediction of preoperative recurrence in non-small cell lung cancer (NSCLC) that is suitable for clinical application is still an open question. Recent advancements integrating genomic data with deep learn...
详细信息
ISBN:
(数字)9798350371499
ISBN:
(纸本)9798350371505
The precise prediction of preoperative recurrence in non-small cell lung cancer (NSCLC) that is suitable for clinical application is still an open question. Recent advancements integrating genomic data with deep learning have shown promise in enhancing recurrence analysis in NSCLC patients. However, the lack of interpretability in the decision-making process of DNN models has hindered their clinical trustworthiness. In this paper, we propose a novel Biologically Informed Pathway-Aware Neural Network (BioPAN). By automatically extracting biological prior knowledge to guide the architecture of DNN models, we design a unified architecture of gene-pathway-biological process-disease. This approach endows each neuron with entity meaning and learns a multi-level view of biological pathways and processes related to recurrence for fully interpretable NSCLC recurrence prediction. We demonstrated that the proposed model well explains the molecular mechanisms linking genes to NSCLC recurrence, identifies several genes that significantly promote and inhibit recurrence, and elucidates the pivotal roles of various gene pathways at the biological process level. Moreover, it outperforms classical machine learning methods, provides fully interpretable biological information, and requires fewer parameters. Broadly, BioPAN enables clinical discovery and prediction, and may have general applicability across cancer types.
The growth of video content in recent years is a challenging problem due to increased memory storage and time consuming for analyzing content of the video. Therefore, there is a need to reduce the content for human us...
详细信息
Emotions have a tremendous impact on human experiences and decision-making processes. People are frequently motivated to repeat happy behaviors while avoiding depressing ones. Emotions may spread like wildfire in the ...
详细信息
ISBN:
(数字)9798350377002
ISBN:
(纸本)9798350377019
Emotions have a tremendous impact on human experiences and decision-making processes. People are frequently motivated to repeat happy behaviors while avoiding depressing ones. Emotions may spread like wildfire in the modern digital era due to the quick broadcast of information and the heavy use of emotive language in communication. The identification of these emotional undercurrents in a variety of literary forms has been made easier by the analysis of indigenous languages, allowing for prompt interventions when needed. This study presents an algorithm that can analyze statements and then categorize them into sentiments in order to determine a person's emotional state. An analysis of the algorithm's performance shows that BERT performs better than SVM and CNN in a variety of emotional categories, showing better recall, accuracy, and F1-scores in the identification of emotions including Joy, Sadness, Fear, and Anger. For example, BERT outperforms SVM and CNN by large margins, achieving an average F1-score of 0.85 across all emotions. CNN performs competitively, especially when it comes to identifying emotions such as Fear and Sadness, but SVM performs somewhat worse overall than BERT and CNN. Interestingly, all models are good at recognizing Neutral emotions, with BERT and CNN doing somewhat better than SVM. As a result, BERT stands out as the best model for classifying emotions, demonstrating its effectiveness in identifying complex emotional states.
Automating the synthesis of User Interfaces (UIs) plays a crucial role in enhancing productivity and accelerating the development lifecycle, reducing both development time and manual effort. Recently, the rapid develo...
详细信息
We propose a novel piecewise stationary linear bandit (PSLB) model, where the environment randomly samples a context from an unknown probability distribution at each changepoint, and the quality of an arm is measured ...
ISBN:
(纸本)9798331314385
We propose a novel piecewise stationary linear bandit (PSLB) model, where the environment randomly samples a context from an unknown probability distribution at each changepoint, and the quality of an arm is measured by its return averaged over all contexts. The contexts and their distribution, as well as the changepoints are unknown to the agent. We design Piecewise-Stationary ε-Best Arm Identification+ (PSεBAI+), an algorithm that is guaranteed to identify an ε-optimal arm with probability ≥ 1 - δ and with a minimal number of samples. PSεBAI+ consists of two subroutines, PSεBAI and NAÏVE ε-BAI(NεBAI), which are executed in parallel. PSεBAI actively detects changepoints and aligns contexts to facilitate the arm identification process. When PSεBAI and NεBAI are utilized judiciously in parallel, PSεBAI+ is shown to have a finite expected sample complexity. By proving a lower bound, we show the expected sample complexity of PSεBAI+ is optimal up to a logarithmic factor. We compare PSεBAI+ to baseline algorithms using numerical experiments which demonstrate its efficiency. Both our analytical and numerical results corroborate that the efficacy of PSεBAI+ is due to the delicate change detection and context alignment procedures embedded in PSεBAI.
Alzheimer's disease (AD) is a widespread neurolog-ical condition affecting millions globally. It gradually advances, leading to memory loss, cognitive deterioration, and a substantial decline in overall quality of...
Alzheimer's disease (AD) is a widespread neurolog-ical condition affecting millions globally. It gradually advances, leading to memory loss, cognitive deterioration, and a substantial decline in overall quality of life for those affected. AD patients experience memory decline, eroding cherished memories and straining relationships, while daily tasks become challenging. Numerous investigations have been conducted in this field, as the timely identification of Alzheimer's disease at its initial stage is of the utmost importance. A major limitation in this field is the predominant emphasis on using single fine-tuned CNN architecture or comparing pre-trained and custom CNN models for Alzheimer's detection, often on small datasets, which neglects a more comprehensive approach. Using smaller datasets can negatively impact deep learning modeling accuracy due to overfitting, limited representation, and poor generalization. This study addresses the current research problems and proposes an ensemble approach that combines predictions from various pre-trained models, including DenseNet-121, EfficientNet B7, ResNet-50, VGG-19, and also from a Custom CNN. The model averaging ensemble method was applied, a subset of the Stacking Ensemble, to two ADNI datasets, with dataset-I being the larger. The goal was to assess the efficacy of this ensemble approach for accurate multiclass classification on ADNI datasets, where it successfully identified all classes despite differing sample volumes. A vast experiment was conducted on two distinct and widely recognized real-world datasets, resulting in accuracies of 99.96% and 98.90% respectively. Finally, the outcome of the research compared with recent research findings demonstrates the potential of our approach in advancing Alzheimer's disease detection by outperforming other benchmark approaches by a significant margin.
Low light region is one of the difficult scenarios for surveillance system, especially noisy, low-quality images may lead to poor object detection and scene analysis. To tackle this problem, we introduce a hybrid GAN ...
详细信息
ISBN:
(数字)9798331509828
ISBN:
(纸本)9798331509835
Low light region is one of the difficult scenarios for surveillance system, especially noisy, low-quality images may lead to poor object detection and scene analysis. To tackle this problem, we introduce a hybrid GAN - CNN framework for low light image denoising and quality enhancement. We develop a framework that combines the strengths of Generative Adversarial Networks (GANs) for photorealistic image generation and Convolutional Neural Networks (CNNs) for robust noise reduction. First denoising and feature extraction is done by the CNN module, and the GAN part further refines the denoised output to improve the perceptual quality of the generated images. We train and validate the proposed model in standard low light image dataset and show superior PSNR, SSIM, and visual clarity compared to state of art methods. We show that the hybrid GAN-CNN model successfully trades between image denoising accuracy and visual realism, thereby providing a viable alternative for enhancing surveillance performance in lowlight conditions. Keywords: Image denoising, GAN-CNN hybrid, low-light enhancement, PSNR, SSIM, surveillance systems.
暂无评论