Lightweight image encryption has become a critical area of cryptography, especially for resource-constrained devices like those used in the Internet of Things (IoT). Chaotic maps, known for their sensitivity to initia...
详细信息
ISBN:
(数字)9798331531492
ISBN:
(纸本)9798331531508
Lightweight image encryption has become a critical area of cryptography, especially for resource-constrained devices like those used in the Internet of Things (IoT). Chaotic maps, known for their sensitivity to initial conditions, ergodicity, and seemingly random behavior, are commonly used in image encryption algorithms. These maps are employed in either a standalone or combined fashion during the diffusion or confusion stages of the encryption process. The high correlation between pixels in image data necessitates specialized algorithms. Pixel scrambling, or confusion, is a fundamental step in image encryption. It involves rearranging pixel positions to disrupt the correlation between neighboring pixel values. This study enhances the Corner Traversal algorithm, a pixel scrambling technique, by using chaotic maps to determine the algorithm parameters at each layer. This modification increases the nonlinearity of the encryption process. The results demonstrate that these improvements lead to a more secure encryption scheme compared to the original Corner Traversal algorithm. The performance of the enhanced algorithm is evaluated by comparing it to the original version.
This paper considers downlink multi-user transmission facilitated by a reconfigurable intelligent surface (RIS). First, focusing on the multi-group multicast beamforming scenario, we develop a fast and scalable algori...
详细信息
Predicting whether the module is faulty is vital to concentrating on the testing and evaluation process. We proposed a machine learning model to predict the faulty model in object-oriented systems. We employ a set of ...
详细信息
ISBN:
(数字)9798331508944
ISBN:
(纸本)9798331508951
Predicting whether the module is faulty is vital to concentrating on the testing and evaluation process. We proposed a machine learning model to predict the faulty model in object-oriented systems. We employ a set of metrics to learn and test the module. The results show averages of 0.82, 0.98, 0.82, and 0.90 for precision, recall, and F1-score, *** proposed model can be used to predict faulty models and decrease the effort of software testing.
Self-Optimizing Memory Controllers present a great potential in the future of memory controllers. As they alleviate the burden of designing an optimal memory scheduling policy, while providing adaptability to differen...
详细信息
Dynamic hedging is a financial strategy that consists in periodically transacting one or multiple financial assets to offset the risk associated with a correlated liability. Deep Reinforcement Learning (DRL) algorithm...
详细信息
The pervasive use of online social media platforms has changed the communication habits and subsequently the content and the frequency of the human exchanges. People share their thoughts on social media more frequentl...
详细信息
ISBN:
(数字)9798350350180
ISBN:
(纸本)9798350350197
The pervasive use of online social media platforms has changed the communication habits and subsequently the content and the frequency of the human exchanges. People share their thoughts on social media more frequently and freely. In particular and due to the societal stigma, people find less obstacle to discuss mental health problems on social media compared to face-o-face discussion. Mining social media for the sake of analysing suicidal intention is an attractive and at the same time challenging task especially during COVID-19 era. The main objective is to leverage machine learning to classify the social media user suicidal behavior while considering the impact of COVID-19 circumstances. We propose a rigorous feature engineering technique based on TF/IDF and Bag of Words. Machine learning classifiers and ensemble models are compared to find that the Neural Network(NN) classifier outperforms the benchmarks with a precision of 94%, a Recall of 94%, an F1 Score of 94% , and an overall accuracy of 94%. In the future, more feature selection techniques and deep learning can be used.
Brain stroke is one of the most common causes of death, ranking as the second leading cause worldwide. With the advancements of deep learning, the detection of brain strokes from CT images becomes possible. In this st...
详细信息
ISBN:
(数字)9798350354133
ISBN:
(纸本)9798350354140
Brain stroke is one of the most common causes of death, ranking as the second leading cause worldwide. With the advancements of deep learning, the detection of brain strokes from CT images becomes possible. In this study, we propose a method for classifying brain stroke images and predicting the presence of a stroke using convolutional neural networks (CNNs), which are particularly effective in image classification tasks. These techniques involve training different CNN models (EfficientNetB0-B7, InceptionResNetV2, ResNet50, ResNet152, and VGG19) on a dataset of labeled brain strokes and normal brain images. Through this process, the model learns to identify patterns and features that are indicative of a stroke. The results show that the different CNN models were able to classify stroke images with high accuracy and predict the presence of stroke with an overall accuracy of 99.95%, which is a significant improvement compared to similar studies. This demonstrates the potential of deep learning-based approaches using CNNs in assisting the diagnosis of strokes and aiding in early-stage treatment. Finally, we provide an analysis of model performance, highlighting each CNN model used, and providing insights for future research.
Fake news continues to proliferate, posing an increasing threat to public discourse. The paper proposes a framework of a Mixture of Experts, Sentiment Analysis, and Sarcasm Detection experts for improved fake news det...
详细信息
ISBN:
(数字)9798331517878
ISBN:
(纸本)9798331517885
Fake news continues to proliferate, posing an increasing threat to public discourse. The paper proposes a framework of a Mixture of Experts, Sentiment Analysis, and Sarcasm Detection experts for improved fake news detection. This approach captures the emotional cues in the text through a Sentiment Analysis expert, which is based on bidirectional encoder representations from Transformers (BERT) models with sentiment vectors generated using SentiWordNet and Integrated Gradients. It combines a sarcasm detection expert based on BERT, recognizing sarcasm and its type to help classify fake news. By fusing these experts through a Mixture of Experts gateway, subtle linguistic cues often found in fake news are more effectively analyzed, leading to improved accuracy in detecting misinformation. Experimental results are presented as 96% for the Sarcasm expert with the BERT base model and 83% for the Sentiment Analysis expert with the distilled version of the BERT (DistilBERT) base model, proving the effectiveness of the proposed approach in beating traditional methods.
This paper studies deep learning (DL)-based energy efficiency maximization (EEM) in multi-simultaneous transmission and reflection-reconfigurable intelligent surfaces (STAR-RISs) assisted massive multiple-input multip...
详细信息
ISBN:
(数字)9798331507022
ISBN:
(纸本)9798331507039
This paper studies deep learning (DL)-based energy efficiency maximization (EEM) in multi-simultaneous transmission and reflection-reconfigurable intelligent surfaces (STAR-RISs) assisted massive multiple-input multiple-output (mMIMO)-non-orthogonal multiple access (NOMA) networks. We formulate the EEM problem to jointly optimize the precoding matrix and STAR-RIS phase shifts subject to the power budget at the base station, STAR-RIS phase shift constraints, and minimum quality-of-service (QoS) requirements. The formulated EEM problem belongs to the mixed-integer programming class, which is difficult to solve optimally. Thus, we develop an alternative optimization approach by dividing the original EEM problem into two sub-problems such as phase shift and beamforming optimization, and solve them alternatively. A bisection search algorithm is proposed to solve the phase shift optimization, while the inner approximation method is employed to address the non-convex beamforming problem through our newly tractable transformations. To enable real-time optimization, we design a DL framework that predicts optimal phase shifts and precoding matrices under various parameter settings. Simulation results demonstrate that the DL-based approach accurately predicts the optimal solutions and is significantly faster than conventional methods. We also evaluate the impact of the essential parameters on the system's performance.
There is a misconception that agile development requires minimal planning effort. In reality, an agile approach for market-driven software development requires highly disciplined, reliable, and accurate planning pract...
详细信息
暂无评论