Human–machine interaction covers a range of applications in which machines should understand humans’ commands and predict their behavior. Humans commonly change their mood over time, which affects the way we interact, particularly by changing speech style and facial expressions. As interaction requires quick decisions, low latency is critical for real-time processing. Edge devices, strategically placed near the data source, minimize processing time, enabling real-time decision-making. Edge computing allows us to process data locally, thus reducing the need to send sensitive information further through the network. Despite the wide adoption of audio-only, video-only, and multimodal emotion recognition systems, there is a research gap in terms of analyzing lightweight models and solving privacy challenges to improve model performance. This motivated us to develop a privacy-preserving, lightweight, CNN-based (CNNs are frequently used for processing audio and video modalities) audiovisual emotion recognition model, deployable on constrained edge devices. The model is further paired with a federated learning protocol to preserve the privacy of local clients on edge devices and improve detection accuracy. The results show that the adoption of federated learning improved classification accuracy by ~2%, as well as that the proposed federated learning-based model provides competitive performance compared to other baseline audiovisual emotion recognition models.

Automatic segmentation of particular heart parts plays an important role in recognition tasks, which is utilized for diagnosis and treatment. One particularly important application is segmentation of epicardial fat (surrounds the heart), which is shown by various studies to indicate risk level for developing various cardiovascular diseases as well as to predict progression of certain diseases. Quantification of epicardial fat from CT images requires advance image segmentation methods. The problem of the state-of-the-art methods for epicardial fat segmentation is their high dependency on user interaction, resulting in low reproducibility of studies and time-consuming analysis. We propose in this paper a novel semiautomatic approach for segmentation and quantification of epicardial fat from 3D CT images. Our method is a semisupervised slice-by-slice segmentation approach based on local adaptive morphology and fuzzy c-means clustering. Additionally, we use a geometric ellipse prior to filter out undesired parts of the target cluster. The validation of the proposed methodology shows good correspondence between the segmentation results and the manual segmentation performed by physicians.

Computer-aided process planning represents the main link between computer-aided design and computer-aided manufacturing. One of the crucial tasks in computer-aided process planning is an operation sequencing problem. In order to find the optimal process plan, operation sequencing problem is formulated as an NP hard combinatorial problem. To solve this problem, a novel genetic crow search approach (GCSA) is proposed in this paper. The traditional CSA is improved by employing genetic strategies such as tournament selection, three-string crossover, shift and resource mutation. Moreover, adaptive crossover and mutation probability coefficients were introduced to improve local and global search abilities of the GCSA. Operation precedence graph is adopted to represent precedence relationships among features and vector representation is used to manipulate the data in the Matlab environment. A new nearest mechanism strategy is added to ensure that elements of machines, tools and tool approach direction (TAD) vectors are integer values. Repair strategy to handle precedence constraints is adopted after initialization and shift mutation steps. Minimization of total production cost is used as the optimization criterion to evaluate process plans. To verify the performance of the GCSA, two case studies with different dimensions are carried out and comparisons with traditional and some modern algorithms from the literature are discussed. The results show that the GCSA performs well for operation sequencing problem in computer-aided process planning.

Gaussian Mixture Models (GMMs) are used in many traditional expert systems and modern artificial intelligence tasks such as automatic speech recognition, image recognition and retrieval, pattern recognition, speaker recognition and verification, financial forecasting applications and others, as simple statistical representations of underlying data. Those representations typically require many high-dimensional GMM components that consume large computing resources and increase computation time. On the other hand, real-time applications require computationally efficient algorithms and for that reason, various GMM similarity measures and dimensionality reduction techniques have been examined to reduce the computational complexity. In this paper, a novel GMM similarity measure is proposed. The measure is based on a recently presented nonlinear geometry-aware dimensionality reduction algorithm for the manifold of Symmetric Positive Definite (SPD) matrices. The algorithm is applied over SPD representations of the original data. The local neighborhood information from the original high-dimensional parameter space is preserved by preserving distance to the local mean. Instead of dealing with high-dimensional parameter space, the method operates on much lower-dimensional space of transformed parameters. Resolving the distance between such representations is reduced to calculating the distance among lower-dimensional matrices. The method was tested within a texture recognition task where superior state-of-the-art performance in terms of the trade-off between recognition accuracy and computational complexity has been achieved in comparison with all baseline GMM similarity measures.