International Journal of Electrical and Computer Engineering Systems

High-Performance Graph Storage and Mutation for Graph Processing and Streaming: A Review

Soukaina Firmli, Dalila Chiadmi — Wed, 18 Dec 2024 00:00:00 +0100

The growing need for managing extensive dynamic datasets has propelled graph processing and streaming to the forefront of the data processing community. Given the irregularity of graph workloads and the large scale of real-world graphs, researchers face numerous challenges when designing high-performance graph processing and streaming systems, due to the sheer volume, intricacy, and continual evolution of graph data. In this paper, we highlight the challenges related to two vital aspects within Graph Processing Systems that significantly impact the overall system performance: 1) the graph storage, encompassing the data structures storing vertices and edges, and 2) graph mutation protocols, referring to the ingestion and storage of new graph updates, such as additions of edges and vertices. Our paper provides a practical taxonomy of techniques designed to improve the efficiency of graph storage and mutation, by reviewing state-of-the-art systems and highlighting the challenges they face in offering a good performance tradeoff for read, write, and memory consumption. Consequently, this enables us to highlight overlooked aspects of performance, that are essential for real-world applications, such as the lack of mutation protocols for graph properties and auxiliary graph data, lack of configurability and cross-platform evaluation of solutions for graph processing and streaming.

Polarization Reconfigurable Patch Antenna Using Parasitic Elements for Sub–6 GHz Applications

Kanniyappan Vinayagam, Rajesh Natarajan — Mon, 09 Dec 2024 00:00:00 +0100

Polarization reconfigurable antenna using parasitic elements are designed for sub-6 GHz applications. A circular patch antenna is designed along with two semicircular arc - elements attached to the radiator with four diodes. By controlling the ON and OFF states of the diodes, the polarization of the antenna can be switched between LHCP and RHCP. Parasitic elements are characterized and placed around the conducting patch to enhance the gain of the antenna. The antenna exhibits a better gain of 5 dBi in both the polarization states. The prototype antenna is fabricated on a FR-4 substrate with full ground plane and tested for reflection coefficient, radiation pattern and polarization conversion ratio. The results are compared with the simulated one and they are having highest correlation between them.

Comparative Analysis of Deepfake Detection Models on Diverse GAN-Generated Images

Mon, 23 Dec 2024 00:00:00 +0100

Advancement in Artificial intelligence has resulted in evolvement of various Deepfake generation methods. This subsequently leads to spread of fake information which needs to be restricted. Deepfake detection methods offer solution to this problem. However, a particular Deepfake detection method which gives best results for a set of Deepfake images (generated by a particular generation method) fails to detect another set of Deepfake images (generated by another method). In this work various Deepfake detection methods were tested for their suitability to decipher Deepfake images generated by various generation methods. We have used VGG16, ResNet50, VGG19, and MobileNetV2 for deepfake detection and pre-trained models of StyleGAN2, StyleGAN3, and ProGAN for fake generation. The training dataset comprised of 200000 images, 50 % of which were real and 50% were fake. The best performing Deepfake detection model was VGG19 with more than 96 percent accuracy for StyleGAN2, StyleGAN3, and ProGAN- generated fakes.

Vector Control of the Induction Motor Based on Whale Optimization Algorithm

Fadhil Hasan — Tue, 10 Dec 2024 00:00:00 +0100

This paper presents the Whale Optimizing Algorithm (WOA) to improve the performance of the induction motors through vector control (VC). The optimization algorithm is utilized to tune the proportional-integral (PI) controllers in both the outer and inner controlling loops. The parameters of these controllers are crucial components of the control system. The WOA is inspired by the social behavior of humpback whales, which is a powerful meta-heuristic algorithm as compared to other techniques. The controlling system and the WOA are implemented using MATLAB-SIMULINK environments. Simulation results demonstrate that this approach significantly improves both dynamic and steady-state responses of the induction motor compared to other optimization techniques. Simultaneously, the success of the WOA in reaching the global optimal parameters can be realized by the significant reduction in computation time and iterations as compared to other methods. The results show a considerable enhancement of about 2% in rise time, 30% in overshoot, and 60% in settling time in accelerating mode in conjunction with a reference case. Also, it gives an improvement of about 9% in rise time, 11% in overshoot, and 64% in settling time in step response. This research contributes to the field of motor control by providing an efficient and reliable optimization method for enhancing the performance of induction motors for various industrial applications.

Time Domain and Qualitative Analysis of a Compact Asymmetrically Fed Circular UWB Antenna for WBAN Scenarios

Venkatesh P, Narmadha T V, Ponnrajakumari M, Lavanya K — Wed, 11 Dec 2024 00:00:00 +0100

This article presents about the design, time domain analysis and qualitative analysis of a circular monopole Ultra Wideband (UWB) antenna for Wireless Body Area Networks (WBAN) applications. The size of the proposed antenna is 30 x 30 x 1 mm3. The proposed antenna provides Ultra wide bandwidth from 2 – 10 GHz and also complies with IEEE C95.3 safety standards. The simulated and measured results are close to each other with minimum deviations. To ensure proper and less distorted communication in real time scenarios, time domain analysis was done for free space, on body and off body conditions. The magnitude and phase of transmission coefficient (S21) were found to be consistent. The group delay was analyzed under free space, on &off body conditions whose variations are less than 0.5 ns in the entire UWB range. Fidelity factor was also analyzed for flat and bent conditions to ensure pulse similarity. Also to ensure the communication link quality, the obtained minimum path loss of 51.10 dB and maximum Received Signal Power of 0.17 dBm were found to be satisfactory warranting a good transmission and reception characteristics of the proposed antenna.

Exploring Speech Emotion Recognition in Tribal Language with Deep Learning Techniques

Wed, 11 Dec 2024 00:00:00 +0100

Emotion is fundamental to interpersonal interactions since it assists mutual understanding. Developing human-computer interactions and a related digital product depends heavily on emotion recognition. Due to the need for human-computer interaction applications, deep learning models for the voice recognition of emotions are an essential area of research. Most speech emotion recognition algorithms are only deployed in European and a few Asian languages. However, for a low-resource tribal language like KUI, the dataset is not available. So, we created the dataset and applied some augmentation techniques to increase the dataset size. Therefore, this study is based on speech emotion recognition using a low-resourced KUI speech dataset, and the results with and without augmentation of the dataset are compared. The dataset is created using a studio platform for better-quality speech data. They are labeled using six perceived emotions: ସଡାଙ୍ଗି (angry), େରହା (happy), ଆଜି (fear), ବିକାଲି (sad), ବିଜାରି (disgust), and େଡ଼କ୍‌(surprise). Mel-frequency cepstral coefficient (MFCC) is used for feature extraction. The deep learning technique is an alternative to the traditional methods to recognize speech emotion. This study uses a hybrid architecture of Long Short-Term Memory (LSTM) and Convolutional Neural Networks (CNNs) as classification techniques for recognition. The results have been compared with existing benchmark models, with the experiments demonstrating that the proposed hybrid model achieved an accuracy of 96% without augmentation and 97% with augmentation.

Augmented Language Dataset for Enhanced Personality Profiling

Mohmad Azhar Teli, Manzoor Ahmad Chachoo — Thu, 02 Jan 2025 00:00:00 +0100

The lexical hypothesis asserts that language encompasses all meaningful individual differences in personality. Language is a vital tool for communication and self-expression, making it essential for understanding and assessing human personality. This paper investigates personality recognition from language use, emphasizing the significance of language in capturing and analyzing personality traits. A comprehensive literature review examines various approaches and techniques in personality recognition. We investigate the effectiveness of language use in predicting personality traits, employing multiple feature extraction and data augmentation techniques to enhance the accuracy and robustness of the personality recognition models. Our approach involves training a generative model, PersonaG, on the Essays dataset, subsequently using it to generate augmented data (AUG-Essays). We compare the performance of machine learning classifiers using LIWC, TF-IDF, Glove, and Word-Vec features on both Essays and AUG-Essays datasets. Our findings demonstrate significant improvements in predictive performance, offering valuable insights for applications in human resources, marketing, and beyond.

Solanaceae Safeguard: CNN-Swin Fusion for Precision Disease Management

Jaferkhan. P, V. Amsaveni — Wed, 18 Dec 2024 00:00:00 +0100

Agricultural productivity stands as a cornerstone of India's economy, and enhancing it remains a priority. A pivotal strategy in bolstering agricultural output is the timely identification of diseases. In agriculture, disease detection and management are crucial for ensuring crop health and yield. This study proposes a novel disease detection system for Solanaceae Vegetables utilizing a hybrid deep learning approach. The system integrates SWIN Transformer architecture with Convolutional Neural Networks (CNN) to analyze and classify disease patterns in Solanaceae vegetables. The dataset used for training and evaluation is sourced from Kaggle repository, comprising comprehensive images of diseased and healthy Solanaceae plants. Through extensive experimentation, the proposed hybrid model achieves a remarkable classification accuracy of 96%. The model demonstrated high precision, recall, and F1-scores across most classes, such as Class 0 (0.92, 0.89, 0.91) and Class 14 (0.97, 1.00, 0.99).The system's high accuracy demonstrates its potential as a reliable tool for disease detection and effective management strategies in Solanaceae vegetable cultivation, thereby contributing to enhanced leaf health and productivity.