Volume 12 | Issue 5 |

 Abstract

Upon completing higher secondary education, students around the globe often find themselves at a crossroads, unsure of which career path to pursue. This pivotal moment demands a level of maturity and self-awareness that many students may not yet possess. As individuals progress through these stages, they inevitably confront the question of what to pursue post-graduation. Our proposed solution tackles this challenge with a computerized career guidance system. By objectively assessing individual skills, this system aims to predict the most suitable career path for each student. Through this process, we aim to provide clarity and direction as students navigate their academic and professional journey.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Counsell Career, Learning Algorithms for machines, Classification(key words)

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

A large number of sanitation workers die every year due to erratic and lack of facilities available, and harmful toxic gases released while cleaning the sewage. This can include monitoring the environment for air quality, temperature, humidity, and sound levels, as well as tracking employee activity and movement. This project aims to develop an innovative Smart Safety Monitoring System (SSMS) for sewage workers, leveraging the capabilities of the Internet of Things (IoT) technology. Real time health monitoring systems for such workers will prove helpful. Sewage workers face numerous risks while performing their duties in confined and hazardous environments. The SSMS is designed to enhance their safety and improve communication. This real time health monitoring device will work in a sewage as a safety equipment. In this project, the device presented will monitor the pulse rate of a person using a pulse oximetry sensor, the methane concentration and the atmospheric oxygen concentration and provide alert to worker and exterior unit. when parameters deviate from the safe range. This parameters in real time will promptly alert the workers to stay safe and detect toxic gases before any harm.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

SMART SAFETY MONITORING SYSTEM FOR SEWAGE WORKERS WITH TWO WAY COMMUNICATION

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

A groundbreaking flood prediction system emerges, blending meteorological, hydrological, and geospatial data with crowd-sourced inputs, all harmonized within a dynamic machine learning structure. Rigorous assessments affirm its prowess, particularly noting the efficacy of a multi-layer perceptron artificial neural network (MLP ANN) setup in delivering precise forecasts. This pioneering methodology harbors promise in bolstering flood mitigation tactics, streamlining preemptive actions, and fortifying rescue endeavors. This cutting-edge approach marks a pivotal advancement in flood management, poised to revolutionize how communities respond to and mitigate the impacts of inundation events.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

INTELLIGENT FLOOD FORECASTING SYSTEM EMPOWERED BY MACHINE LEARNING

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Cardiovascular diseases, particularly heart disease, remain a leading cause of mortality worldwide, necessitating advanced diagnostic systems that leverage clinical data for early and accurate prediction. Machine Learning integration techniques, particularly ensemble methods, is a way that enhances the precision and reliability of predictive models for heart disease diagnosis. The complex nature of heart diseases demands a comprehensive analysis of clinical data to derive actionable insights. While traditional diagnostic approaches have relied on individual risk factors, the combination of diverse clinical parameters offers a more broad perspective, enabling a more understanding and prediction of cardiovascular outcomes.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Data Classification, Learning Algorithms for machines, Data Analysis

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Yoga pose estimation is a computer vision technique used to predict the position/ pose of a part of the human body. This paper presents the framework to analyse and assess yoga postures by designing, developing, and implementing a Yoga Posture Detection System using computer vision and deep learning (DL) models such as CNN and VGG16. The study employs advanced image processing algorithms to extract information from images or videos of individuals performing yoga poses. We employed several postures which includes camel, downdog, goddess, plank, tree, and warrior2. With the use of a deep learning model that has been built, the system is able to precisely recognise and categorise different positions while providing instantaneous feedback on proper alignment, balance, and posture.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

CONVOLUTIONAL YOGA POSE ESTIMATOR

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

To label this, a paper has been initiated to create a machine learning model for Speech emotion recognition (SER) involves the identification of emotions conveyed in spoken language through analysis of speech signals. With the growing popularity of voice assistants and smart speakers, SER has gained significant attention in recent years. One approach to SER is to use deep learning models such as "Convolutional Neural Networks " (CNNs) and Long Short-Term Memory (LSTM) networks. In this particular paper, we suggest a novel approach for SER using a 2D CNN- LSTM architecture. The proposed model first uses a 2D CNN to extract the relevant characterstics from the speech signal, followed by a LSTM network for sequence modeling. We evaluated our proposed model on the Berlin Emotional Speech Database (EMO-DB), achieving state-of-the-art results. We also balance our model's performance with other existing SER models and found that our suggested model outperformed them. Our speculative results shows that the proposed 2D CNN-LSTM architecture is an effective method for SER and can be used in real-world applications such as recognition of emotion from voice assistants, call centers, and customer service applications.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

CNN, LSTM, 2D CNN LSTM.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

This paper conducts a comparative analysis of four reinforcement learning (RL) algorithms--Q-learning, Deep Q-Networks (DQN), SARSA, and Proximal Policy Optimization (PPO)--using the Pong game as a benchmark. Each algorithm's performance, convergence rate, and computational efficiency are evaluated. Results indicate that while Q-learning and SARSA exhibit simplicity, they struggle with discrete action spaces. DQN, with its capability to handle continuous action spaces, shows improved performance but requires longer training times. PPO demonstrates a balance between sample efficiency and computational complexity, achieving faster convergence and superior performance. This Examination sheds light on selecting appropriate RL algorithms for real-world applications.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Reinforcement Learning, Pong Game, Q-learning, DQN, SARSA, PPO, Comparative Analysis, Performance Evaluation, Convergence Rate, Computational Efficiency.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

In recent years, transformer-based models have reshaped the landscape of computer visions, particularly in image classification tasks Vision Transformers (ViT) and BEiT (BERT Pre-Training of Image Transformers) stand out as notable examples, employing self-attention mechanisms. This paper presents a detailed comparative analysis of ViT and BEiT, aiming to elucidate their performance, strengths, weaknesses, and interpretability in image classification Through extensive experimentation across diverse benchmark datasets like CIFAR-10, CIFAR-100, and ImageNet[1], we evaluate the models based on classification accuracy, training efficiency, generalization capability, and robustness to adversarial perturbations Our findings offer insights at nuanced differences between ViT and BEiT, revealing ViT's efficiency and small-scale datasets, while highlighting BEiT's enhanced robustness to adversarial attacks and domain shifts Furthermore, we research the interpretability of learned representations and visualize attention patterns generated. The ability to capture meaningful image features and the comparative analysis not merely informs practitioners and researchers in computer visions but also paves the way for further advancements in transformer-based architectures for visual understanding.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Transformer-based Models, Vision Transformers, BEiT, Image Classification, Self-Attention Mechanisms, Comparative Analysis, Interpretability, Robustness, Adversarial Attacks, Computer Vision.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

This paper introduces an innovative fusion of classical dance and artificial intelligence, focusing on the esteemed art form of Bharatanatyam. Our pioneering framework harnesses the power of Generative AI techniques to revolutionize both the creation and experience of Bharatanatyam performances. Through advanced machine learning models, textual descriptions are seamlessly translated into visually captivating dance sequences, effectively capturing the essence and intricacies of this ancient art form. The system not only facilitates the creation of choreography but also offers a user-friendly interface tailored for artists, enthusiasts, and learners alike, thereby fostering unprecedented engagement with Bharatanatyam. By meticulously preserving the grammatical structure and predefined steps inherent in Bharatanatyam, our approach ensures an authentic representation of this rich cultural heritage. Moreover, this project serves as a catalyst for revitalizing classical dance by infusing it with cutting-edge technology, while simultaneously encouraging creative exploration and interpretation. We firmly believe that this harmonious convergence of tradition and technology will not only redefine the boundaries of artistic expression but also significantly impact the future trajectory of cultural preservation and appreciation.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Bharatanatyam, Generative AI, Dance Creation, Cultural Heritage, Artistic Expression.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

This paper presents a comprehensive framework for combating fake news by integrating deepfake video detection and text analysis techniques. With the proliferation of misinformation, especially through deepfake technology, there is an urgent need for robust detection methods. Our approach involves extracting text from social media posts, generating interrogative sentences, querying a web server for relevant information, and summarizing the authenticity of news, videos, or posts. By combining advanced AI algorithms for deepfake detection and text analysis, our framework offers a powerful solution to enhance the credibility of news sources and combat the spread of misinformation in digital media. Keywords-- Deepfake video detection, Text analysis, Fake news detection, Misinformation, Artificial intelligence (AI), Generative AI, Deep learning (DL), Natural language processing (NLP), social media, Web server querying, Factchecking, Digital media ecology.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Deepfake video and Text Detection, LSTM, Artificial intelligence (AI), Generative AI, Deep learning (DL), Natural language processing (NLP).

  License

Creative Commons Attribution 4.0 and The Open Definition