A5035097654- Blind Source Separation Techniques
- Spectroscopy and Chemometric Analyses
- Advanced Adaptive Filtering Techniques
- Speech and Audio Processing
- Wireless Communication Networks Research
- Educational Robotics and Engineering
- Image and Signal Denoising Methods
- AI in cancer detection
- Digital Imaging for Blood Diseases
- Direction-of-Arrival Estimation Techniques
- Quantum Computing Algorithms and Architecture
- Cell Image Analysis Techniques
- Neural dynamics and brain function
- Fractal and DNA sequence analysis
- Sensory Analysis and Statistical Methods
- Quantum-Dot Cellular Automata
- Advanced Clustering Algorithms Research
- Sparse and Compressive Sensing Techniques
- Quantum Information and Cryptography
- EEG and Brain-Computer Interfaces
- Neural Networks and Applications
- Cloud Computing and Resource Management
- Medical Image Segmentation Techniques
- Emotion and Mood Recognition
- Bayesian Methods and Mixture Models
Universidad de Sevilla
2010-2024
Imperial College London
2013
Ayuntamiento de Sevilla
2004
This paper demonstrates the integration of Reinforcement Learning (RL) into quantum transpiling workflows, significantly enhancing synthesis and routing circuits. By employing RL, we achieve near-optimal Linear Function, Clifford, Permutation circuits, up to 9, 11 65 qubits respectively, while being compatible with native device instruction sets connectivity constraints, orders magnitude faster than optimization methods such as SAT solvers. We also significant reductions in two-qubit gate...
The problem of blind source separation complex-valued sources from a linear mixture is addressed. We propose deflationary algorithm for the sequential recovery set communication signals, where each extracted by performing Bounded Component Analysis mixture. contribution recovered to observations removed minimizing its convex perimeter, without using second-order statistics. This implies run gradient descent several times. In order accelerate convergence, we have derived fast step size that...
Centroid-based clustering is a widely used technique within unsupervised learning algorithms in many research fields. The success of any centroid-based relies on the choice similarity measure under use. In recent years, most studies focused including several divergence measures traditional hard k-means algorithm. this article, we consider problem using family α β -divergences, which governed by two parameters, and . We propose new iterative algorithm, -k-means, giving closed-form solutions...
Immunohistochemistry is a powerful technique that widely used in biomedical research and clinics; it allows one to determine the expression levels of some proteins interest tissue samples using color intensity due biomarkers with specific antibodies. As such, immunohistochemical images are complex their features difficult quantify. Recently, we proposed novel method, including first separation stage based on non-negative matrix factorization (NMF), achieved good results. However, this method...
In this paper, we propose a method for solving the permutation problem that is inherent in separation of convolved mixtures speech signals time-frequency domain. The proposed obtains solution through maximization contrast function exploits similarity temporal envelope spectrum. For purpose, calculation uses global measure based on recently developed family generalized Alpha-Beta divergences, which depend two tuning parameters, alpha and beta. This parameterization exploited to best spectrum...
This paper studies the problem of blind extraction a subset bounded component signals from observations linear mixture. In first part this paper, we analyze geometric assumptions that characterize problem, and their implications on mixing matrix latent sources. second part, solve by adopting principle minimizing risk, which refers to encoding complexity in worst admissible situation. provides an underlying justification several analysis (BCA) criteria, including minimum normalized volume...
The blind decomposition of the observations, as a set additive components simpler structure, is problem with many applications in scientific and practical fields. Our study assumes that component signals are bounded nature, relies on geometric convex supports observations Minkowski direct sum sets support components. This last property, which weaker than mutual independence sufficient for essential identifiability indecomposable In practice, it usual lie one-dimensional complex subspaces....
This article addresses the problem of unsupervised separation speech signals in realistic scenarios. An initialization procedure is proposed for independent component analysis (ICA) algorithms that work time-frequency domain and require prewhitening observations. It shown method drastically reduces permuted solutions helps to reduce execution time algorithms. Simulations confirm these advantages several ICA instantaneous effectiveness technique emulated reverberant environments.
Blind separation of speech sources in reverberant environments is usually performed the time-frequency domain, which gives rise to permutation problem: different ordering estimated for frequency components. A two-stage method solve permutations with an arbitrary number proposed. The suggested procedure based on spectral consistency sources. At first stage bins are compared each other, while at second neighboring frequencies emphasized. Experiments perfect situations and live recordings show...
In many research laboratories, it is essential to determine the relative expression levels of some proteins interest in tissue samples. The semi-quantitative scoring a set images consists establishing scale scores ranging from zero or one maximum number by researcher and assigning score each image that should represent predefined characteristic IHC staining, such as its intensity. However, manual depends on judgment an observer therefore exposes assessment certain level bias. this work, we...
This paper addresses the problem of blind detection a desired user in an asynchronous DS-CDMA communications system with multipath propagation channels. Starting from inverse filter criterion introduced by Tugnait and Li 2001, we propose to tackle context signal extraction methods for ICA. In order improve performance detector, present based on joint optimization several higher-order statistics outputs. An algorithm that optimizes proposed is described, its improved robustness respect...
Immunohistochemistry is a powerful technique, widely used in biomedical research and clinics, that allows to determine the expression levels of some proteins interest tissue samples by means intensity color due biomarkers using specific antibodies. As such, immunohistochemical images are complex problematic be quantified. Recently we proposed novel method including first separation stage based on nonnegative matrix factorization (NMF) achieved good results. However, was highly dependent...
Automatic emotion recognition systems aim to identify human emotions from physiological signals, voice, facial expression or even physical activity.Among these types of the usefulness signals electroencephalography (EEG) should be highlighted.However, there are few publicly accessible EEG databases in which induction is performed through virtual reality (VR) scenarios.Recent studies have shown that VR has great potential evoke an effective and natural way within a laboratory environment.This...
This paper addresses the problem of supervised signal extraction a desired in linear mixture. The main assumption is bounded support sources, property that allows to propose loss function based on convex perimeter training error. work reviews our recent results Bounded Component Analysis error with special emphasis harder case where mixture underdetermined, is, there are more sources than sensors. In this scenario, due lack degrees freedom, one cannot extract perfectly source even noiseless...