New algorithms improve signal processing in federated learning.
problem Efficiently process distributed signal samples with privacy and communication constraints.
method Proposes overpredictive signal approximations using convex optimization.
result Quantifies tradeoffs between communication cost, sampling rate, and approximation error.
Automated feature recommendation for signal analytics with interpretability.
problem Signal data analytics problems requiring feature interpretation.
method Wide Learning architecture for feature recommendation with interpretability.
result Effective feature recommendation and interpretation techniques for signal analytics.
Semi-supervised learning method augments minority class examples for robust anomaly detection in clinical signals.
problem Class imbalance in minority class instances impairs robustness of clinical analytics solutions.
method Intelligent augmentation of minority class examples to balance class distribution and construct a smooth decision boundary.
result The proposed method outperforms state-of-the-art algorithms in anomaly detection for clinical signals.
Study on signal recovery from low-rank matrix with sparse noise.
problem Inference of a rank-one signal in the presence of sparse noise.
method Replica method from statistical physics, recursive distributional equations, population dynamics algorithm.
result Critical signal strength for recovery via top eigenvector identified.
A new method for joint noise removal and trend estimation from sparse signals.
problem Jointly removing noise and estimating trends from sparse signals.
method PENDANTSS combines SOOT/SPOQ penalties with BEADS algorithm in a Trust-Region block alternating variable metric forward-backward approach.
result Outperforms comparable methods in deconvolving analytical chemistry signals.
A challenging problem in physics concerns the possibility of forecasting rare but extreme phenomena such as large earthquakes, financial market crashes, and material rupture. A promising line of research involves the early detection of precursory log-periodic oscillations to help forecast extreme events in collective p…
We give a new, very general, formulation of the compressed sensing problem in terms of coordinate projections of an analytic variety, and derive sufficient sampling rates for signal reconstruction. Our bounds are linear in the coherence of the signal space, a geometric parameter independent of the specific signal and m…
Field theory explains optimal scaling in ResNets for signal propagation.
problem Understanding optimal scaling parameter for ResNet performance.
method Finite-size field theory for ResNets to study signal propagation and scaling.
result Analytical expressions for optimal scaling parameter, independent of other hyperparameters.
New algorithm improves signal recovery from noisy measurements with theoretical guarantees.
problem Recovering signals from noisy measurements in inverse problems.
method Wasserstein-based projections (WP) replacing analytic regularization with data-driven denoising.
result WP approximates true projection with high probability, providing theoretical guarantees.
Survey of tensor completion algorithms for big data analytics.
problem Filling missing entries in tensors.
method Overview of recent tensor completion algorithms.
result Advances in tensor completion for big data.
A new method for accurately reconstructing signals without knowing the kernel or signal regularity.
problem Recovering signals from noisy measurements without prior knowledge of the convolution kernel or signal regularity.
method Parametrizing the convolution kernel and prior length-scales, jointly estimated in the inversion procedure.
result Accurate reconstructions of signals with varying regularity and unknown kernel size.
Missing data reduces signal-to-noise ratio, not sample size, for PCA.
problem Effect of missing data on PCA signal structure learning.
method Analytic and simulation studies of probabilistic PCA with missing data.
result Missing data effectively reduces signal-to-noise ratio, not sample size.
There are three equivalent ways of representing two jointly observed real-valued signals: as a bivariate vector signal, as a single complex-valued signal, or as two analytic signals known as the rotary components. Each representation has unique advantages depending on the system of interest and the application goals. I…
New neural network extracts signal components and their IFs from non-uniform samples.
problem Recovering signal components and their IFs from discrete blind-source data.
method Inspired by theory, deep neural network extends Hilbert transform and synchrosqueezed wavelet transform.
result Neural network resolves inverse problem for non-uniformly sampled data.
This research uses machine learning to approximate ideal and hotelling observer performance for binary signal detection.
problem Approximating the Ideal and Hotelling Observers for binary signal detection tasks.
method Supervised learning methods, including CNNs and SLNNs, are employed to approximate the IO and HO test statistics.
result The proposed supervised learning methods provide accurate approximations of the IO and HO test statistics.
Optimal trading strategy adapts to signals in markets with price impact.
problem Optimal liquidation in markets with linear price impact and predictive signals.
method Formulated as a stochastic control problem, solved using probabilistic and convex analytic techniques.
result Explicit solution for optimal trading strategy in terms of SDEs.
This paper explores NMF identifiability and its applications.
problem Understanding NMF identifiability for better interpretability and applications.
method Comprehensive tutorial on NMF identifiability, connections to algorithms and applications.
result Significant progress in NMF identifiability research since the 2010s.
Survey visual analytics methods for detecting anomalous user behaviors.
problem Understanding and detecting anomalous user behaviors in various domains.
method Survey and classification of visual analytics methods in four categories.
result Discussion of findings and potential research directions.
Paper explores supervised learning methods to approximate ideal observer for joint signal detection and localization.
problem Optimizing medical imaging systems by assessing their performance using the Ideal Observer model.
method Uses supervised learning methods, specifically convolutional neural networks, to approximate the Ideal Observer for joint signal detection and localization tasks.
result Supervised learning-based methods can approximate the Ideal Observer for joint signal detection and localization tasks, as shown by comparisons to MCMC and analytical methods.
We investigate the presence of residual multifractal background for monofractal signals which appears due to the finite length of the signals and (or) due to the long memory the signals reveal. This phenomenon is investigated numerically within the multifractal detrended fluctuation analysis (MF-DFA) for artificially g…
The paper analyzes MACD using operator theory.
problem Understanding the mathematical foundation of MACD.
method Developed a functional-analytic framework interpreting MACD as a phase-corrected, smoothed derivative operator.
result MACD is structurally equivalent to a band-pass filter and can be expressed as a finite difference of delayed and doubly averaged signals.
New method recovers signals from saturated data using linear loss and nonconvex penalties.
problem Signal recovery from saturated measurements with sign information loss.
method Linear loss and nonconvex penalties (e.g., minimax concave penalty, sorted ℓ1 norm).
result Estimation error is bounded and recovery performance improved.
ARMA graph filters approximate any graph frequency response and are stable in time-varying settings.
problem Designing efficient graph filters for signals on graphs.
method Autoregressive moving average (ARMA) recursions for graph filtering.
result ARMA graph filters can approximate any desired graph frequency response and are stable in time-varying settings.
Proposes a probabilistic framework for stationary topological signals on simplicial complexes.
problem Complex data structures require new models and tools.
method Generalizes stationarity to topological signals on simplicial complexes.
result Defines topological power spectral density (PSD) for stationary signals.
Bayesian compressive sensing speeds up object detection in video sequences.
problem Efficiently detecting objects in large video datasets.
method Bayesian compressive sensing methods for object detection.
result Bayesian methods achieve similar or better accuracy than greedy algorithms but faster.
Many techniques in computer vision, machine learning, and statistics rely on the fact that a signal of interest admits a sparse representation over some dictionary. Dictionaries are either available analytically, or can be learned from a suitable training set. While analytic dictionaries permit to capture the global st…
Proposes DNN for enhancing sound quality scores.
problem Improving sound quality assessment scores.
method Develops a DNN optimization scheme based on black-box optimization and policy gradient method.
result Significant increase in OSQA scores without minimizing MSE.
Paper proposes algorithms for robust 1-bit compressive sensing with nonconvex penalties.
problem Recovering sparse signals from one-bit measurements.
method Develops algorithms based on convex and nonconvex penalties, providing analytical solutions.
result Analytical solutions for several nonconvex penalties are found, making the recovery process faster and more efficient.
Method separates target signal properties from noisy mixtures.
problem Signal recovery from noisy mixtures with specific statistical properties.
method Statistical component separation method using noise samples and matching statistics.
result Method outperforms standard denoising methods in recovering target signal properties.
Develops algorithms for sparse signal reconstruction without needing signal sparsity or noise variance.
problem Sparse signal reconstruction challenges due to unknown signal sparsity and noise variance.
method TF-IGP and RRT-IGP frameworks for OMP and OLS without prior knowledge of k0 and σ2. result TF-IGP and RRT-IGP achieve successful sparse recovery under restricted isometry conditions.
Paper revisits five IF paradoxes using differential geometry.
problem Five paradoxes of Instantaneous Frequency in three-phase systems.
method Geometric interpretation of frequency to explain IF paradoxes.
result Revisits and explains five IF paradoxes through a common framework.
New techniques for compressive sensing without noise or signal statistics.
problem Support recovery in underdetermined linear regression models without prior noise and signal statistics.
method Proposes RRM and RRTA to operate OMP algorithm without noise variance or signal sparsity knowledge.
result Establishes high SNR consistency for OMP without prior noise and signal statistics.
This tutorial covers methods for handling missing data in SP and ML.
problem Dealing with missing data in signal processing and machine learning.
method Grouping strategies into three tasks: imputation, estimation, and prediction.
result Promising and future research directions are discussed.
The paper introduces a multi-scale model to speed up sparse approximation problems for visual signals.
problem Efficiently solving sparse approximation problems with large dictionaries.
method Incorporates multi-scale structure onto dictionary-based sparse representations.
result Significant speedups (10-60x) with little loss in accuracy for images, videos, and light fields.
Paper develops DL methods for signal demodulation in wireless comms.
problem Signal demodulation in wireless communications.
method Proposes DBN-SVM and AdaBoost demodulators using real modulated signals.
result Proposed DBN-SVM and AdaBoost demodulators outperform traditional methods.
Optimal liquidation strategy with price impact and signal exploitation.
problem Maximizing revenue-risk in a market with transient and temporary price impact.
method Infinite dimensional stochastic control approach, backward stochastic differential equation, operator-valued Riccati equation.
result Explicit expression for the optimal trading strategy.
Robust algorithm identifies sparse signals using L2 regularization.
problem Reconstructing sparse signals from noisy data using overcomplete dictionaries.
method Corrected Projections Algorithm (CPA) with L2 regularization.
result CPA efficiently identifies known atoms in noisy signals.
Proposes SVR-based image denoising using natural image relations.
problem Image denoising with arbitrary noise sources.
method Support Vector Regression (SVR) in wavelet domain, enforcing natural image relations.
result Outperforms conventional methods and is similar to state-of-the-art methods for Gaussian noise.
Investor optimizes stock investments with noisy future price signals.
problem Optimizing stock investments with uncertain future stock prices.
method Dynamic investment strategy with partial observation of Brownian motion.
result Closed-form solution for optimal investment problem.
Bayesian model improves spectral estimation from partial, noisy data.
problem Challenges in spectral estimation with partial and noisy observations.
method Joint probabilistic model with Gaussian process prior and Bayes' rule for exact inference.
result Proposed model provides functional-form representation of power spectral density.
Develops algorithm to infer correlated signals from unknown structures.
problem Inference of correlated signal fields with unknown correlation structures.
method Free energy exploration (FrEE) strategy within information field theory (IFT).
result Algorithm efficiently identifies optimal field estimates and their uncertainties.
Big data analytics improves healthcare through early detection and quality life.
problem Limited access to healthcare data hinders evidence-based decision-making.
method Analysis of healthcare data using various tools and techniques.
result Big data analytics enhances healthcare quality and patient outcomes.
Study on infinite neural networks for generalization.
problem Understanding generalization in infinitely-wide neural networks.
method Analytical and empirical investigations of infinite ensembles.
result Signals correlating with generalization identified.
Proposes random Euler filters for efficient complex-valued nonlinear signal processing.
problem Efficiently processing complex-valued nonlinear signals with reduced computational cost.
method Introduces linear and widely-linear random Euler complex-valued filters with fixed network structures.
result Analytical minimum mean square error and optimum step-size derived for transient and steady-state performances.
This paper uses social signals to improve cryptocurrency price forecasting.
problem Improving cryptocurrency price forecasting using social signals.
method LSTMs trained on historical price data and social data from GitHub and Reddit.
result Social signals reduce error in forecasting cryptocurrency prices, especially for Bitcoin.
New method improves signal reconstruction with nonconvex penalties and parameter control.
problem Reconstructing sparse signals with nonconvex penalties and nonconvexity control.
method Introduces nonconvex penalties (SCAD, MCP) with nonconvexity parameters and controls them to guide AMP trajectory.
result Achieves perfect reconstruction for relatively dense signals with small nonconvexity parameters.
Paper tackles edge computing for reliable ITS data analytics.
problem Real-time data analytics for ITSs with low latency and reliability.
method Edge-centric architecture with deep learning for heterogeneous data.
result Preliminary results show reliable, secure ITS environment.
A new algorithm for high-dimensional hedging problems.
problem High-dimensional, path-dependent hedging problems.
method Signature-based algorithm using operator-valued kernels and geometric rough paths.
result Theoretical guarantees on existence and uniqueness of a global minimum.