The paper "Cyberbullying Detection" by Roshni Jadhav, Grisha Chaudhari, and Sumeet Rane from the Information Technology Department at Atharva College of Engineering in Mumbai, India, addresses the growing issue of cyberbullying, which has become a worldwide epidemic. The authors highlight the negative physical and emotional impacts of cyberbullying and the need for effective detection methods. They review existing techniques, noting that while many studies have focused on specific social media platforms or topics, deep learning models have shown promise in overcoming these limitations by leveraging transfer learning. The paper discusses the use of deep neural networks, specifically Convolutional Neural Networks (CNNs), Long Short-Term Memory (LSTMs), and Bidirectional LSTMs (BLSTMs), for detecting cyberbullying. These models are compared in terms of their performance and adaptability across different datasets. The authors re-implement these models using a Twitter dataset to evaluate their interoperability and performance on new platforms. The proposed system integrates word embedding techniques and deep learning models to enhance the detection of cyberbullying posts. The system's effectiveness is demonstrated through experimental results, showing high accuracy in identifying both non-bullying and bullying comments. The paper concludes that the proposed system is a robust solution for detecting cyberbullying, offering valuable insights for reducing its prevalence on social media platforms.The paper "Cyberbullying Detection" by Roshni Jadhav, Grisha Chaudhari, and Sumeet Rane from the Information Technology Department at Atharva College of Engineering in Mumbai, India, addresses the growing issue of cyberbullying, which has become a worldwide epidemic. The authors highlight the negative physical and emotional impacts of cyberbullying and the need for effective detection methods. They review existing techniques, noting that while many studies have focused on specific social media platforms or topics, deep learning models have shown promise in overcoming these limitations by leveraging transfer learning. The paper discusses the use of deep neural networks, specifically Convolutional Neural Networks (CNNs), Long Short-Term Memory (LSTMs), and Bidirectional LSTMs (BLSTMs), for detecting cyberbullying. These models are compared in terms of their performance and adaptability across different datasets. The authors re-implement these models using a Twitter dataset to evaluate their interoperability and performance on new platforms. The proposed system integrates word embedding techniques and deep learning models to enhance the detection of cyberbullying posts. The system's effectiveness is demonstrated through experimental results, showing high accuracy in identifying both non-bullying and bullying comments. The paper concludes that the proposed system is a robust solution for detecting cyberbullying, offering valuable insights for reducing its prevalence on social media platforms.