CICIDS-2017 Dataset Feature Analysis with Information Gain for Anomaly Detection Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 32 No. 1s (2025) 182 https://internationalpubls.com Integrating Diverse Data Streams for Enhanced Emotional Intelligence in Mental Health Care Asha Priyadarshini.M1, Dr. A. Krishna Mohan2 1Research Scholar, JNTUK, Kakinada. 2Professor, Dept. of CSE, University College of Engineering Kakinada, JNTUK, Kakinada, AP, India. ashapriyadarshini16@gmail.com Article History: Received: 11-08-2024 Revised: 26-09-2024 Accepted: 13-10-2024 Abstract: This research presents a comprehensive system for real-time emotion recognition and analysis using multimodal data, including image, video, audio, and text. The system employs deep learning models to extract features and classify emotions from each modality. By integrating these predictions, we aim to provide a holistic understanding of a user's emotional state and assess potential risks. The system further analyzes the collected emotion data to identify trends, patterns, and indicators of emotional well- being and suicide risk. Visualizations such as time-series plots, distribution charts, and stacked area charts are employed to represent the emotional dynamics over time.Based on the analysis; the system generates personalized recommendations and alerts for users exhibiting signs of distress or emotional instability, including potential suicide risk factors. The goal is to promote proactive well-being management and early intervention. It is crucial to emphasize that this system is a tool for early detection and should be used in conjunction with professional mental health care. Keywords: multimodal emotion recognition, convolutional neural networks, recurrent neural networks, facial expression analysis, speech emotion recognition, well-being assessment, suicide risk prediction, feature extraction, data fusion, natural language processing I. INTRODUCTION Mental health has emerged as a critical global concern, underscored by the increasing prevalence of emotional disorders and the devastating impact of suicide. Traditional methods of mental health assessment often rely on subjective self-reports, which can be prone to biases and inaccuracies. Advancements in artificial intelligence and computer vision have opened new avenues for objective and continuous monitoring of emotional states. This research aims to develop an innovative system for real-time emotion recognition and analysis using multimodal data, including image, video, audio, and text. By leveraging the power of deep learning, the system seeks to extract meaningful insights from these diverse data sources to gain a comprehensive understanding of a user's emotional state. Human emotions are complex and multifaceted, influenced by various factors such as facial expressions, body language, vocal tone, and linguistic patterns. By combining these modalities, we aim to create a more robust and accurate emotion recognition system. The integration of these Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 32 No. 1s (2025) 183 https://internationalpubls.com modalities allows for complementary information extraction and can mitigate the limitations of relying on a single modality. Early detection of emotional distress and potential suicide risk is crucial in preventing tragic outcomes. This system is designed to go beyond basic emotion recognition by analyzing emotional patterns over time. By tracking changes in emotional states, identifying trends, and detecting anomalies, we aim to provide early warning signals of potential mental health crises. The system employs advanced deep learning models to process and analyze multimodal data. Convolutional neural networks (CNNs) are utilized for image and video analysis, while recurrent neural networks (RNNs) and transformer-based models are employed for audio and text processing, respectively. These models are trained on large datasets to achieve high accuracy in emotion classification. A key component of the system is the development of robust feature extraction techniques. For image and video data, facial landmarks and facial action units are extracted to capture subtle emotional cues. For audio data, acoustic features such as mel-frequency cepstral coefficients (MFCCs) and pitch are extracted to represent vocal characteristics. For text data, natural language processing techniques are used to extract semantic and syntactic information. The extracted features are then fed into the respective deep learning models to predict emotions. The system incorporates a fusion module to combine the predictions from different modalities, resulting in a more comprehensive and reliable emotion assessment. By continuously monitoring a user's emotional state, the system can generate valuable insights into their well-being. Visualizations and analytics tools are integrated to provide users with a clear understanding of their emotional patterns and trends. Additionally, the system can generate personalized recommendations based on the detected emotional states, such as stress management techniques, relaxation exercises, or seeking professional help. It is important to emphasize that this system is not intended to replace professional mental health care. Instead, it aims to complement existing approaches by providing an additional layer of monitoring and support. The system should be used as a tool for early detection and intervention, encouraging users to seek professional help when necessary. By combining advanced machine learning techniques with multimodal data analysis, this project seeks to contribute to the development of innovative solutions for mental health care and well-being. II. LITERATURE SURVEY: A substantial body of research has explored emotion recognition across various modalities, with a focus on text, speech, and visual data. Early works in text-based emotion recognition primarily employed rule-based or lexicon-based approaches [11]. However, recent advancements in deep learning have led to the development of more sophisticated models. For instance, Rajabi, Shehu, and Uzuner et al. [12] proposed a multi-channel BiLSTM-CNN model for multi-label emotion classification, demonstrating improved performance over traditional methods. Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 32 No. 1s (2025) 184 https://internationalpubls.com In the realm of speech emotion recognition, researchers have explored various deep learning architectures. Zhao, Zhao, and Tian [22] introduced a multiscale deep convolutional LSTM network to effectively capture temporal and spectral features of speech signals. Similarly, Wang et al. [23] proposed a dual-sequence LSTM architecture for improved emotion classification. These studies highlight the potential of deep learning in accurately recognizing emotions from speech data. Visual emotion recognition has also witnessed significant progress, with CNN-based models achieving promising results. Zhang, Yangsen, et al. [16] proposed a coordinated CNN-LSTM- attention model for text sentiment classification, demonstrating the effectiveness of combining different neural network architectures. Park, Bae, and Cheong [15] introduced an emotion embedding model for recognizing emotions from text stories. While individual modalities have been extensively studied, research on multimodal emotion recognition is still in its early stages. Integrating information from multiple sources has the potential to enhance emotion recognition accuracy and robustness. However, challenges such as data fusion and computational complexity need to be addressed. This review highlights the growing interest in emotion recognition and the potential of deep learning techniques in this field. By combining insights from these studies, this research aims to develop a robust multimodal emotion recognition system capable of accurately classifying emotions and providing valuable insights into user well-being. III. METHODOLOGY Data Acquisition and Preprocessing Multimodal Data Collection: The system collects multimodal data, including image, video, audio, and text, from diverse sources. Image and video data are captured through cameras or user-provided media. Audio data is obtained through microphones or audio recordings. Textual data can be derived from user inputs, social media posts, or other relevant sources. Data Preprocessing: Prior to analysis, the collected data undergoes rigorous preprocessing. Image and video data are subjected to face detection and facial landmark extraction to isolate relevant regions of interest. Audio data is segmented into overlapping frames and converted into spectrograms for feature extraction. Textual data is cleaned, tokenized, and converted into numerical representations suitable for machine learning models. Feature Extraction Image and Video Features: To capture facial expressions and body language, we extract features such as facial landmarks, facial action units (FAUs), and optical flow. These features are essential for recognizing emotions like happiness, sadness, anger, and fear. Audio Features: Mel-Frequency Cepstral Coefficients (MFCCs), pitch, and intensity are extracted from audio signals to represent vocal characteristics associated with different emotions. Textual Features: Natural Language Processing (NLP) techniques are employed to extract semantic and syntactic features from textual data. Word embeddings, sentiment analysis, and topic modelling are utilized to capture the emotional content of the text. Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 32 No. 1s (2025) 185 https://internationalpubls.com Model Development and Training Deep Learning Architectures: • Image and Video: Convolutional Neural Networks (CNNs) are employed for feature extraction and classification of facial expressions and body language. • Audio: Recurrent Neural Networks (RNNs) or Convolutional Neural Networks (CNNs) are used for processing audio spectrograms and classifying emotions based on vocal characteristics. • Text: Recurrent Neural Networks (RNNs), Long Short-Term Memory (LSTM), or Transformer-based models are utilized for text classification, capturing the emotional context of the text. Model Training: The models are trained on large-scale datasets containing labeled emotional data. Transfer learning is employed to leverage pre-trained models and improve performance. Data augmentation techniques are used to enhance model robustness and generalization. Multimodal Fusion: To combine the predictions from different modalities, a fusion approach is adopted. Early fusion, late fusion, or a hybrid approach can be explored based on the specific application and performance evaluation. Emotion Recognition and Analysis The trained models are applied to real-time or stored data to predict emotions. The system continuously monitors the emotional state of the user by analyzing the incoming multimodal data. Emotion labels are generated for each modality, and these predictions are fused to obtain a final emotion prediction. Time-series analysis is conducted on the predicted emotions to identify patterns, trends, and anomalies. Statistical methods and machine learning algorithms are employed to detect significant changes in emotional states. Well-being Assessment and Risk Prediction Based on the analyzed emotional data, a well-being score is calculated. This score reflects the overall emotional state of the user and can be used to monitor changes in well-being over time. To assess suicide risk, additional features such as social isolation, sleep patterns, and behavioral changes can be incorporated. Machine learning models are trained to predict the likelihood of suicide risk based on these features and the emotional state of the user. A. Novelty of the Research The novelty of this research lies in its comprehensive and integrated approach to emotion recognition and analysis for well-being and suicide risk assessment. While previous research has explored individual modalities (image, audio, text) for emotion recognition, this research distinguishes itself by combining multiple modalities to create a more robust and accurate system. Moreover, the focus on real-time analysis and continuous monitoring of emotional states is a significant advancement over traditional methods. By leveraging deep learning techniques and Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 32 No. 1s (2025) 186 https://internationalpubls.com advanced feature extraction methods, the research aims to achieve higher accuracy and sensitivity in emotion detection compared to existing approaches. The incorporation of well-being assessment and suicide risk prediction based on the analysis of emotional patterns is another novel aspect of this research. By identifying trends, anomalies, and potential risk factors, the system offers a proactive approach to mental health care. Additionally, the development of personalized recommendations based on individual emotional profiles represents a significant contribution to the field of mental health support. Overall, this research novelty stems from its multimodality, real-time analysis, focus on well-being and suicide risk assessment, and the integration of personalized recommendations. B. Dataset Analysis and Description Image and Video Data The foundation of our image and video emotion recognition model is the FER2013 dataset, a widely recognized benchmark containing approximately 30,000 facial images with seven basic emotion labels (angry, disgust, fear, happy, sad, surprise, neutral). To enhance the model's robustness and generalization, we incorporated additional datasets such as RAF-DB and AffectNet. These datasets provide a more diverse range of facial expressions, including complex emotions and variations in lighting and pose. Audio Data For audio-based emotion recognition, we utilized the RAVDESS dataset, which comprises acted emotional speech audio files with variations in emotional intensity and vocal pitch. To augment the dataset and improve model performance, we included the CREMA-D and IEMOCAP datasets, which offer a wider range of emotional expressions and real-world speech scenarios. Text Data To evaluate the text-based emotion recognition component, we employed the Restaurant Reviews dataset. This dataset contains a large volume of text data with associated sentiment labels, providing a suitable benchmark for assessing the model's ability to classify emotions from textual content. While this dataset primarily focuses on sentiment analysis, it can be adapted to emotion classification by mapping sentiment labels to corresponding emotional categories. Note: For optimal performance and generalization, data preprocessing techniques such as data augmentation, normalization, and feature scaling were applied to all datasets. C. Algorithm Justifications: Emotion Recognition: Image and Video Modality: • Convolutional Neural Networks (CNNs): To extract discriminative features from facial images and videos, we employed CNN architectures such as VGG, ResNet, or EfficientNet. Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 32 No. 1s (2025) 187 https://internationalpubls.com These models were fine-tuned on the aforementioned datasets to achieve optimal performance in emotion classification. • Transfer Learning: To leverage pre-trained models and accelerate training, transfer learning was employed. Pre-trained models like VGG, ResNet, and Inception were fine-tuned on the target datasets. Audio Modality: • Mel-Frequency Cepstral Coefficients (MFCCs): To extract relevant features from audio signals, MFCCs were computed to represent the spectral content of the audio. • Deep Neural Networks (DNNs): DNNs, including Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs), were employed to classify emotions based on the extracted MFCC features. Text Modality: • Natural Language Processing (NLP): Textual data was preprocessed using techniques like tokenization, stemming, and stop word removal. • Recurrent Neural Networks (RNNs): Long Short-Term Memory (LSTM) networks or Gated Recurrent Units (GRUs) were used to capture the sequential nature of text and classify emotions based on textual content. Multimodal Fusion • Early Fusion: Feature-level fusion was explored by concatenating the extracted features from different modalities before feeding them into a single classifier. • Late Fusion: Decision-level fusion was implemented by combining the predictions from individual modality classifiers using techniques like weighted averaging or majority voting. Well-being Assessment and Suicide Risk Prediction • Machine Learning: To assess well-being and predict suicide risk, machine learning algorithms such as Support Vector Machines (SVMs), Random Forests, or Gradient Boosting were employed. These models were trained on features extracted from the multimodal data, including emotional states, behavioral patterns, and demographic information. Evaluation Metrics • Accuracy, precision, recall, and F1-score: These metrics were used to evaluate the performance of the emotion recognition models. • Confusion matrices: To analyze the classification errors and identify potential biases. • Area Under the Curve (AUC): To assess the performance of the suicide risk prediction model. By combining these algorithms and techniques, the system effectively extracts meaningful information from multimodal data, accurately predicts emotions, and provides valuable insights into user well-being and potential risks. Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 32 No. 1s (2025) 188 https://internationalpubls.com IV. ARCHITECHTURE DESCRIPTION The proposed system architecture comprises several interconnected modules designed for multimodal emotion recognition, analysis, and well-being assessment. Data Acquisition and Preprocessing Module This module serves as the foundation for the system, responsible for collecting and preparing data from various sources. Image and video data are captured through cameras or user-provided media, while audio data is obtained from microphones or audio recordings. Textual data can be derived from user inputs or external sources. Preprocessing steps include face detection, facial landmark extraction for image and video data, segmentation and feature extraction for audio data, and tokenization and cleaning for textual data. Feature Extraction Module Extracting relevant features is crucial for accurate emotion recognition. The system employs a combination of hand-crafted and deep learning-based feature extraction techniques. For image and video modalities, features like facial landmarks, facial action units, and optical flow are computed. Audio features include Mel-Frequency Cepstral Coefficients (MFCCs), pitch, and intensity. Textual features are derived using techniques like word embeddings, sentiment analysis, and topic modelling. Emotion Recognition Module The core of the system, this module utilizes deep learning models to classify emotions based on the extracted features. Convolutional Neural Networks (CNNs) are employed for image and video data, while Recurrent Neural Networks (RNNs) or Convolutional Neural Networks (CNNs) are utilized for audio data. For text data, Recurrent Neural Networks (RNNs) or Transformer-based models are employed. The models are trained on extensive datasets to achieve robust emotion classification. Multimodal Fusion Module To enhance the accuracy and robustness of emotion recognition, a multimodal fusion strategy is adopted. The system integrates information from different modalities to provide a comprehensive understanding of the user's emotional state. Both early fusion (feature-level) and late fusion (decision-level) approaches are explored to determine the optimal fusion strategy. Emotion Analysis and Well-being Assessment Module This module processes the predicted emotions to extract meaningful insights into the user's well- being. Time-series analysis is conducted to identify patterns, trends, and anomalies in emotional states. Statistical methods are employed to calculate metrics such as average emotion levels and well-being scores. Additionally, the system incorporates machine learning models to predict potential suicide risks based on emotional patterns and other relevant factors. User Interface Module A user-friendly interface is provided to visualize the analysis results, present personalized recommendations, and facilitate user interaction. The interface displays emotion trends, distribution charts, and other relevant visualizations. It also offers suggestions for improving well-being based on Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 32 No. 1s (2025) 189 https://internationalpubls.com the detected emotional patterns. The system architecture emphasizes a modular design, allowing for flexibility and scalability. By integrating multiple modalities and leveraging advanced machine learning techniques, the system aims to provide a comprehensive and accurate assessment of user emotions and well-being. Fig 2: Overall Architecture-Flow Diagram V. RESULTS Image and Video Emotion Recognition The image and video emotion recognition model was trained for 60 epochs, achieving a final training accuracy of 98.39% and a loss of 0.0562. The model exhibited strong generalization performance, with a testing accuracy of 98.20%. The training and validation curves indicated stable convergence without signs of overfitting. Speech Emotion Recognition The speech emotion recognition model was trained for 50 epochs, reaching a training accuracy of 96.73% and a validation accuracy of 96.34%. The model demonstrated consistent performance on the testing dataset, achieving an accuracy of 96.34%. The training and validation loss curves exhibited a decreasing trend, indicating effective learning. Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 32 No. 1s (2025) 190 https://internationalpubls.com Text Emotion Recognition The text emotion recognition model was trained for 25 epochs, achieving a final training accuracy of 98.39%, precision of 98.47%, and recall of 98.26%. The model's performance on the validation set was slightly lower, with an accuracy of 93.11%, precision of 94.05%, and recall of 92.81%. On the testing set, the model achieved an accuracy of 94.09%, precision of 95.20%, and recall of 93.39%. The confusion matrix revealed potential misclassifications between certain emotion categories, which could be addressed through further model refinement or data augmentation. Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 32 No. 1s (2025) 191 https://internationalpubls.com Model Outputs: Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 32 No. 1s (2025) 192 https://internationalpubls.com VI. CONCLUSION This research presents a novel approach to emotion recognition and analysis by integrating multimodal data, including image, video, audio, and text. The proposed system successfully employs deep learning models to extract relevant features and classify emotions with high accuracy. The integration of these modalities significantly enhances the system's ability to capture the complexity of human emotions. The analysis of emotional patterns over time provides valuable insights into user well-being. By monitoring changes in emotional states, the system can identify potential signs of distress or emotional instability. The incorporation of suicide risk prediction based on emotional indicators is a significant step towards early intervention and prevention. The experimental results demonstrate the effectiveness of the proposed system in accurately recognizing emotions across different modalities. The models exhibited strong performance in terms of accuracy, precision, and recall, indicating their potential for real-world applications. While the Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 32 No. 1s (2025) 193 https://internationalpubls.com system shows promising results, further research is needed to refine the models and expand the dataset to improve generalization and robustness. Future Scope Building upon the success of this research, future work will focus on several key areas: Enhancing Model Performance: Investigating advanced deep learning architectures and exploring transfer learning techniques to improve the accuracy and efficiency of emotion recognition models. Real-time Implementation: Developing a real-time system capable of processing multimodal data streams efficiently for immediate emotion detection and analysis. Longitudinal Studies: Conducting longitudinal studies to assess the system's ability to track changes in emotional states over extended periods and its effectiveness in predicting long-term well-being outcomes. Personalized Recommendations: Developing more sophisticated recommendation systems tailored to individual users based on their emotional profiles and preferences. Ethical Considerations: Addressing ethical challenges related to privacy, data security, and bias in emotion recognition systems. Clinical Validation: Collaborating with mental health professionals to evaluate the system's clinical utility and its potential impact on mental health care. By addressing these areas, future research can contribute to the development of robust and reliable emotion recognition systems with significant implications for mental health monitoring and intervention. References [1] Kessler, Ronald C., et al. "Suicide prediction models: a critical review of recent research with recommendations for the way forward." Molecular psychiatry 25.1 (2020): 168-179. [2] Belsher, Bradley E., et al. "Prediction models for suicide attempts and deaths: a systematic review and simulation." JAMA psychiatry 76.6 (2019): 642-651. [3] Large, Matthew Michael. "The role of prediction in suicide prevention." Dialogues in clinical neuroscience 20.3 (2018): 197-205. [4] Pokorny, Alex D. "Suicide prediction revisited." Suicide and Life‐Threatening Behavior 23.1 (1993): 1-10. [5] Mann, J. John, et al. "Suicide prevention strategies: a systematic review." Jama 294.16 (2005): 2064-2074. [6] Zalsman, Gil, et al. "Suicide prevention strategies revisited: 10-year systematic review." The Lancet Psychiatry 3.7 (2016): 646-659. [7] Wu, Kevin Chien-Chang, Ying-Yeh Chen, and Paul SF Yip. "Suicide methods in Asia: implications in suicide prevention." International journal of environmental research and public health 9.4 (2012): 1135-1158. [8] Goldney, Robert D. Suicide prevention. Oxford University Press, USA, 2013. [9] Turecki, Gustavo, et al. "Suicide and suicide risk." Nature reviews Disease primers 5.1 (2019): 74. [10] Hawton, Keith. "Assessment of suicide risk." The British Journal of Psychiatry 150.2 (1987): 145-153. [11] Batbaatar, Erdenebileg, Meijing Li, and Keun Ho Ryu. "Semantic-emotion neural network for emotion recognition from text." IEEE access 7 (2019): 111866-111878. [12] Rajabi, Zahra, Amarda Shehu, and Ozlem Uzuner. "A multi-channel bilstm-cnn model for multilabel emotion classification of informal text." 2020 IEEE 14th International Conference on Semantic Computing (ICSC). IEEE, 2020. Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 32 No. 1s (2025) 194 https://internationalpubls.com [13] Kim, Hannah, and Young-SeobJeong. "Sentiment classification using convolutional neural networks." Applied Sciences 9.11 (2019): 2347. [14] Bharti, Santosh Kumar, et al. "Text‐Based Emotion Recognition Using Deep Learning Approach." Computational Intelligence and Neuroscience 2022.1 (2022): 2645381. [15] Park, Seo-Hui, Byung-Chull Bae, and Yun-Gyung Cheong. "Emotion recognition from text stories using an emotion embedding model." 2020 IEEE international conference on big data and smart computing (BigComp). IEEE, 2020. [16] Zhang, Yangsen, et al. "A text sentiment classification modeling method based on coordinated CNN‐LSTM‐attention model." Chinese Journal of Electronics 28.1 (2019): 120-126. [17] Zhang, Yuxiang, et al. "Text Emotion Distribution Learning via Multi-Task Convolutional Neural Network." IJCAI. 2018. [18] Abdullah, Malak, MirsadHadzikadicy, and Samira Shaikhz. "SEDAT: sentiment and emotion detection in Arabic text using CNN-LSTM deep learning." 2018 17th IEEE international conference on machine learning and applications (ICMLA). IEEE, 2018. [19] Xu, Dongliang, et al. "Deep learning based emotion analysis of microblog texts." Information Fusion 64 (2020): 1- 11. [20] Ameer, Iqra, et al. "Multi-label emotion classification in texts using transfer learning." Expert Systems with Applications 213 (2023): 118534. [21] Zhao, Jianfeng, Xia Mao, and Lijiang Chen. "Speech emotion recognition using deep 1D & 2D CNN LSTM networks." Biomedical signal processing and control 47 (2019): 312-323. [22] Zhang, Shiqing, Xiaoming Zhao, and Qi Tian. "Spontaneous speech emotion recognition using multiscale deep convolutional LSTM." IEEE Transactions on Affective Computing 13.2 (2019): 680-688. [23] Wang, Jianyou, et al. "Speech emotion recognition with dual-sequence LSTM architecture." ICASSP 2020-2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 2020. [24] Etienne, Caroline, et al. "Cnn+ lstm architecture for speech emotion recognition with data augmentation." arXiv preprint arXiv:1802.05630 (2018). [25] Ahmed, Md Rayhan, et al. "An ensemble 1D-CNN-LSTM-GRU model with data augmentation for speech emotion recognition." Expert Systems with Applications 218 (2023): 119633. [26] Issa, Dias, M. Fatih Demirci, and Adnan Yazici. "Speech emotion recognition with deep convolutional neural networks." Biomedical Signal Processing and Control 59 (2020): 101894. [27] Patro, Pramoda, et al. "A hybrid approach estimates the real-time health state of a bearing by accelerated degradation tests, Machine learning." 2021 Second International Conference on Smart Technologies in Computing, Electrical and Electronics (ICSTCEE). IEEE, 2021 [28] Wu, Xixin, et al. "Speech emotion recognition using capsule networks." ICASSP 2019-2019 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 2019. [29] Senthilkumar, N., et al. "Speech emotion recognition based on Bi-directional LSTM architecture and deep belief networks." Materials Today: Proceedings 57 (2022): 2180-2184. [30] Dangol, Ranjana, et al. "Speech emotion recognition UsingConvolutional neural network and long-short TermMemory." Multimedia Tools and Applications 79.43 (2020): 32917-32934. [31] Asha Priyadarshini Manda, "Max 30100/30102 sensor implementation to viral infection detection based on Spo2 and heartbeat pattern." Annals of the Romanian Society for Cell Biology (2021): 2053-2061. [32] M. Asha Priyadarshini, “A Comprehensive Analysis of Machine Learning and Deep Learning Approaches Towards IOT Security” IEEE explorer, May,2023 DOI: 979-8-3503-9737-6/23/$31.00 ISBN:979-8-3503-0009-3 [33] K.Venkateswara Rao, ”Suicide Prediction on Social Media by Implementing Sentimental Analysis along with Machine Learning”, International Journal of Recent Technology and Engineering(IJRTE), ISSN : 2277-3878, Vol-8 Issue-2, July 2019, Page No: 4833-4837. [34] Priyadarshini, M. Asha, et al. "A Visionary Approach to Anemia Detection: Integrating Eye Condition Data and Machine Learning." International Conference on Computational Innovations and Emerging Trends (ICCIET-2024). Atlantis Press, 2024. Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 32 No. 1s (2025) 195 https://internationalpubls.com [35] Priyadarshini, M. Asha, et al. "A Data Mining Approach to Monitor Terrorism Dissemination Online." International Conference on Computational Innovations and Emerging Trends (ICCIET-2024). Atlantis Press, 2024. [36] Priyadarshini, M. Asha, et al. "A Multi-Feature Approach with Data Augmentation for Speech Emotion Recognition using Deep Learning." International Conference on Computational Innovations and Emerging Trends (ICCIET- 2024). Atlantis Press, 2024. [37] Salma, S., Priyadarshini, M. A., Manaswini, P. S., Kumar, P. S., Prathyusha, P., & Ganesh, S. (2024, July). Agro- Insight: Recommendation System Using Machine Learning. In International Conference on Computational Innovations and Emerging Trends (ICCIET-2024) (pp. 824-834). Atlantis Press.