The analysis and manipulation of signals, including digital signal processing, image processing, audio processing, compression, filtering, and applications in communications and multimedia
6 papersThe Informer model uses a ProbSparse self-attention mechanism and a distilling operation to efficiently handle long time series sequences, improving both prediction accuracy and computational efficiency. It significantly outperforms traditional and state-of-the-art deep learning models on multiple datasets, especially for longer-term predictions.
This paper surveys the security and privacy landscape of the metaverse, discussing the convergence of technologies, potential threats (e.g., identity theft, data breaches, and physical safety risks), and countermeasures in various application domains (gaming, social experience, online collaboration, and creator economy). It highlights the unique challenges posed by metaverse characteristics like immersiveness, spatiotemporality, and decentralization, emphasizing the need for innovative security solutions in this rapidly evolving space.
This paper provides a tutorial on Principal Component Analysis (PCA) and its variations, including approximate PCA for large datasets and kernel PCA for nonlinear data. It discusses the mathematical foundations of these methods and presents algorithms for their implementation, focusing on reducing computational costs in different scenarios like small sample size, large datasets, and high-dimensional spaces.
This paper introduces a probabilistic grid-based approach for robot mapping in dynamic environments, using cell-specific Hidden Markov Models to represent occupancy and its changes. The method learns state transition probabilities from observed data, enabling online adaptation and prediction of future occupancy states, improving map accuracy and path planning compared to traditional occupancy grids, particularly in semi-static environments like parking lots.
The study found that the choice of similarity measure can significantly impact the results of trajectory analysis. Spatial differences between trajectories were found to be more influential than temporal differences in determining similarity. The paper provides recommendations for selecting and using trajectory similarity measures in practice.
This tutorial introduces Generalized Eigendecomposition (GED), a powerful method for isolating patterns in multichannel electrophysiology data. GED creates spatial filters that maximize researcher-specified contrasts (e.g., different frequency bands or experimental conditions), facilitating denoising, contrast enhancement, and dimension reduction for improved analysis of brain signals.