My research is related to statistical learning, mainly kernel machines and neural networks, for Earth Observation and remote sensing data analysis.
I'm interested in understanding human vision from information theoretic principles. This statistical view has implications in experimental and computational neuroscience. (See the ex-cathedra statement)
My interests are related to machine learning and signal and image processing. The application domains are remote sensing data analysis and hyperspectral images with special focus on cloud screening.
At present I'm focused on kernel methods, support vector machines, semi-supervised and active learning. The main application field is on remote sensing data. I have recently worked with one-class classifiers applied to hyperspectral images.
My current research involves signal processing to develop new interactive man-machine interfaces and haptics, and real-time multispectral sensors and actuators. I'm also with Analog Devices Inc.
I'm working on image fusion, developing downscaling and pansharpening methods. The goal is to tackle multitemporal image fusion and change detection problems with improved resolution.
My research activity is centered in the retrieval of soil moisture and vegetation biogeophysical parameters from space observations (microwave radiometers, radars and hyperspectral imagers), and the development of multi-sensor downscaling techniques for enhanced information.
I am conducting fMRI recordings of the visual brain using synthetic and natural images. I am director of the
Optometry Clinic of the Universitat de Valencia
, which has a range of experimental tools for vision research.
José J. Esteve-Taboada
I am developing computational models of the visual function based on functional Magnetic Resonance Imaging, psychophysics, and image statistics.
I'm working in image statistics and vision science. I have developed several methods for density estimation, measure independence, manifold learning and visual quality assessment.
My background is on metric learning and kernel machines. I'm currently a postdoc at the ISP, working on feature extraction and classification problems in remote sensing cloud detection.
José Enrique Adsuara
Currently, I am working on learning parameters of differential equations, HPC for linear solvers, and causality. I am interested in machine learning, especially statistical learning and deep neural networks.
Ana B. Ruescas
I am a data scientist and project manager. As a remote sensing specialist I have experience in several application areas like ocean colour and thermal algorithm development and validation.
My PhD work included kernel-based nonlinear generalization of classical (linear) feature extraction techniques to improve classification results in remote sensing.
My background is in the field of applied machine learning. I am specially interested in applications to natural sciences like remote sensing and weather forecasting. Currently I am working with convolutional neural networks and kernel methods applied to cloud detection.
Anna Maria Mateo
I am working on developing machine learning algorithms for Earth Observation global monitoring. My current research is the incorporation of physical knowledge and multivariate output regression methods.
Daniel Heestermans Svendsen
I am working on machine learning methods for remote sensing and earth observation data. My current focus is on kernel methods, and the incorporation of physical knowledge in statistical methods.
Jose Antonio Padrón
My PhD thesis work is about developing a new family of anomaly change detection algorithms for remote sensing image processing and geoscience time series analysis.
Juan Emmanuel Johnson
My background is in enhancing manifold learning and alignment algorithms for improved dimensionality reduction and data fusion of remote sensing hyperspectral images. My current research involves feature learning and dependence estimation using kernel methods and semi-supervised learning for applications in Earth observation data.
My research interests include kernel methods, graphical models and causality. I'm interested in applications for detecting deforestation and coral bleaching.
The main goal of my research is to develop automatic algorithms for the detection of clouds from remote sensing images. I mainly focus on the inclusion of prior knowledge and (spatial, temporal, angular) constraints in machine learning classifiers.
My research work is about machine learning and its application on remote sensing specifically in feature extraction and climate dynamics analysis. I am especially interested in soil moisture data analysis and climate teleconnections.
I am working on convolutional versions of linear + nonlinear models of visual neuroscience and using those in visual prosthesis and image quality metrics.
Alumni and past visitors
At IPL I've been working on computational visual neuroscience, modelling the processes that take place in the brain from the retinal images until we get information out of them.
My primary research interests lie in the area of Monte Carlo methods for Bayesian inference. My specialty is focused on the random number generation problem and computational methods for stochastic quadrature, such as rejection sampling, MCMC algorithms and importance sampling techniques.
My current research interests are related to statistical methods for the detection and the attribution of climate change and especially for the attribution of extreme weather events. Attribution methods usually rely on the analysis of observations and climate model experiments.
I'm with the UVERS group, doing my PhD thesis on biophysical parameter retrieval for crop monitoring, in particular in the FP7 ERMES project, and collaborating with ISP people on GP retrieval algorithms.
In my postdoc at the ISP group, I addressed a number of machine learning problems related to hyperspectral image processing including graph adaptation, active learning, and advanced kernel methods.
In my PhD I applied advanced contrast perception models as regularization functionals to solve inverse problems such as image restoration and motion estimation and studied their connection to image statistics.
My PhD work (best-thesis award in Physics and Maths 2003) was focused on perceptual and statistical image representations for image coding and texture classification.
My work at the ISP group included the development of multiinformation and divergence measures using Gaussianization transforms.
In my MSc work I applied accurate contrast perception models to improve Support Vector Regression in subjective domains for image coding.
In my MSc thesis I worked with a Kernel generalization of the SSIM image quality index well suited to be applied to hyperspectral images.
In my PhD years (best-MSc thesis award in Computer Science 2013) I analyzed the complexity of spatio-spectral signals for illumination invariant Bayesian reflectance estimation and hyperspectral image coding.
During my stay at ISP I used with Kernel Ridge Regression for image denoising assuming smoothness in the spatial domain.
In my MSc thesis I applied nonlinear models of chromatic contrast perception in wavelet domains to improve JPEG2000.
During my MSc work I contributed to develop nonlinear local-to-global Independent Component Analysis.
My work at the ISP group included the analysis of multi-temporal remote sensing image changes, and the definition of advanced one-class classifiers.
My work at the ISP group included the development of semi-supervised one-class classifiers for remote sensing data classification.
My work at the ISP group included the development of target detection algorithms for remote sensing data analysis.
At ISP I built virtual worlds of controlled spatial arrangement to study the effects of occlusion, perspective and view point in 2D shape statistics.
in my MSc Thesis I developed in cortical image representations that are simultaneously robust to neural noise and energy efficient.
I'm working in the statistical analysis of fMRI data to develop new models in computational visual neuroscience.
In my stay at IPL I addressed the problem of decoding the visual signals from simulated and real neural responses.