Speech emotion recognition (SER) is the process of predicting human emotions from audio signals using artificial intelligence (AI) techniques. SER technologies have a wide range of applications in areas such as psychology, medicine, education, and entertainment. Extracting relevant features from audio signals is a crucial task in the SER process to correctly identify emotions. Several studies on SER have employed short-time features such as Mel frequency cepstral coefficients (MFCCs), due to their efficiency in capturing the periodic nature of audio signals. However, these features are limited in their ability to correctly identify emotion representations. To solve this issue, this research combined MFCCs and time-domain features (MFCCT) to enhance the performance of SER systems. The proposed hybrid features were given to a convolutional neural network (CNN) to build the SER model. The hybrid MFCCT features together with CNN outperformed both MFCCs and time-domain (t-domain) features on the Emo-DB, SAVEE, and RAVDESS datasets by achieving an accuracy of 97%, 93%, and 92% respectively. Additionally, CNN achieved better performance compared to the machine learning (ML) classifiers that were recently used in SER. The proposed features have the potential to be widely utilized to several types of SER datasets for identifying emotions.

The swift development of the Internet of Things (IoT) has accelerated digitalization across several industries, offering networked applications in fields such as security, home automation, logistics, and quality control. The growth of connected devices, on the other hand, raises worries about data breaches and security hazards. Because of IoT devices’ computational and energy limits, traditional cryptographic methods face issues. In this context, we emphasize the importance of our contribution to image encryption in IoT environments through the proposal of Multiple Map Chaos Based Image Encryption (MMCBIE), a novel method that leverages the power of multiple chaotic maps. MMCBIE uses multiple chaotic maps to construct a strong encryption framework that considers the inherent features of digital images. Our proposed method, MMCBIE, distinguishes itself by integrating multiple chaotic maps like Henon Chaotic Transform and 2D-Logistic Chaotic Transform in a novel combination, a unique approach that sets it apart from existing schemes. Compared to other chaotic-based encryption systems, this feature renders them practically indistinguishable from pure visual noise. Security evaluations and cryptanalysis confirm MMCBIE’s high-level security properties, indicating its superiority over existing image encryption techniques. MMCBIE demonstrated superior performance with NPCR (Number of Pixel Changing Rate) score of 99.603, UACI (Unified Average Changing Intensity) score of 32.8828, MSE (Mean Square Error) score of 6625.4198, RMSE (Root Mean Square Error) score of 80.0063, PSNR (Peak Signal to Noise Ratio) score of 10.2114, and other security analyses.

Internet of Things (IoT) technologies will interconnect with a wide range of network devices, regardless of their local network and resource capacities. Ensuring the security, communication, and privacy protection of end-users is a major concern in IoT development. Secure communication is a significant requirement for various applications, especially when communication devices have limited resources. The emergence of IoT also necessitates the use of low-power devices that interconnect with each other for essential processing. These devices are expected to handle large amounts of monitoring and control data while having limited capabilities and resources. The algorithm used for secure encryption should protect vulnerable devices. Conventional encryption methods such as RSA or AES are computationally expensive and require large amounts of memory, which can adversely affect device performance. Simplistic encryption techniques are easily compromised. To address these challenges, an effective and secure lightweight cryptographic process is proposed for computer devices. This process utilizes a symmetrical encryption key block, incorporating a custom proxy network (SP) and a modified Feistel architecture. Security analysis and performance evaluation results demonstrate that the proposed protocol is secure and energy-efficient. The symmetric key encryption scheme is based on sequences in the Feistel cipher, with multiple rounds and sub-keys generated using principles derived from genetic algorithms. This proposed algorithm minimizes processing cycles while providing sufficient security.

JMIR Human Factors - a leading peer-reviewed ehealth journal

Lifelogging is the process of digital tracking of person’s daily experiences for a variety of purposes. In recent years, lifelogging has become an increasingly popular area of research due to not only the growing demands from many applications, such as wellbeing, entertainment, healthcare systems, and intelligent environments, but also to the advances in device technologies that offer the promise to record and store large volumes of personal data using inexpensive tools. However, getting insights from egocentric experience using huge deluge of unlabeled and unstructured data continues to pose major challenges. A large number of research have been conducted in recent years to cover these challenges but there is still a lack of studies that provide a comprehensive survey of the available literature, while most of the existing lifelogging surveys generally focus on only one aspect. This review highlights the advances of state-of-the-art in lifelogging from different angles, including its research history, current applications, activity recognition techniques, moment retrieval, storytelling, privacy and security issues, as well as challenges and future research trends.

Universities around the world are keen to develop study plans that will guide their graduates to success in the job market. The internship course is one of the most significant courses that provides an experiential opportunity for students to apply knowledge and to prepare to start a professional career. However, internships do not guarantee employability, especially when a graduate’s internship performance is not satisfactory and the internship requirements are not met. Many factors contribute to this issue making the prediction of employability an important challenge for researchers in the higher education field. In this paper, our aim is to introduce an effective method to predict student employability based on context and using Gradient Boosting classifiers. Our contributions consist of harnessing the power of gradient boosting algorithms to perform context-aware employability status prediction processes. Student employability prediction relies on identifying the most predictive features impacting the hiring opportunity of graduates. Hence, we define two context models, which are the student context based on the student features and the internship context based on internship features. Experiments are conducted using three gradient boosting classifiers: e X treme G radient B oosting (XGBoost), C ategory B oosting (CatBoost) and L ight G radient B oosted M achine (LGBM). The results obtained showed that applying LGBM classifiers over the internship context performs the best compared to student context. Therefore, this study provides strong evidence that the employability of graduates is predictable from the internship context.