Ángela Fernández

Hybrid machine learning forecasting of solar radiation values

The constant expansion of solar energy has made the accurate forecasting of radiation an important issue. In this work we apply Support Vector Regression (SVR), Gradient Boosted Regression (GBR), Random Forest Regression (RFR) as well as a hybrid method to combine them to downscale and improve 3-h accumulated radiation forecasts provided by Numerical Weather Prediction (NWP) systems for seven locations in Spain. We use either direct 3-h aggregated radiation forecasts or we build first global accumulated daily predictions and disaggregate them into 3-h values, with both approaches outperforming the base NWP forecasts. We also show how to disaggregate the 3-h forecasts into hourly values using interpolation based on clear sky (CS) theoretical and experimental radiation models, with the disaggregated forecasts again being better than the base NWP ones and where empirical CS interpolation yields the best results. Besides providing ample background on a problem that offers many opportunities to the Machine Learning (ML) community, our study shows that ML methods or, more generally, hybrid artificial intelligence systems are quite effective and, hence, relevant for solar radiation prediction.

Support Vector Forecasting of Solar Radiation Values

The increasing importance of solar energy has made the accurate forecasting of radiation an important issue. In this work we apply Support Vector Regression to downscale and improve 3-hour accumulated radiation forecasts for two locations in Spain. We use either direct 3-hour SVR-refined forecasts or we build first global accumulated daily predictions and disaggregate them into 3-hour values, with both approaches outperforming the base forecasts. We also interpolate the 3-hour forecasts into hourly values using a clear sky radiation model. Here again the disaggregated SVR forecast perform better than the base ones, but the SVR advantage is now less marked. This may be because of the clear sky assumption made for interpolation not being adequate for cloudy days or because of the underlying clear sky model not being adequate enough. In any case, our study shows that machine learning methods or, more generally, hybrid artificial intelligence systems are quite relevant for solar energy prediction.

Diffusion Maps for dimensionality reduction and visualization of meteorological data

The growing interest in big data problems implies the need for unsupervised methods for data visualization and dimensionality reduction. Diffusion Maps (DM) is a recent technique that can capture the lower dimensional geometric structure underlying the sample patterns in a way which can be made to be independent of the sampling distribution. Moreover, DM allows us to define an embedding whose Euclidean metric relates to the sample?s intrinsic one which, in turn, enables a principled application of k-means clustering. In this work we give a self-contained review of DM and discuss two methods to compute the DM embedding coordinates to new out-of-sample data. Then, we will apply them on two meteorological data problems that involve time and spatial compression of numerical weather forecasts and show how DM is capable to, first, greatly reduce the initial dimension while still capturing relevant information in the original data and, also, how the sample-derived DM embedding coordinates can be extended to new patterns.

Diffusion methods for aligning medical datasets: location prediction in CT scan images.

The purpose of this study is to introduce diffusion methods as a tool to label CT scan images according to their position in the human body. A comparative study of different methods based on a k-NN search is carried out and we propose a new, simple and efficient way of applying diffusion techniques that is able to give better location forecasts than methods that can be considered the current state-of-the-art.

On-line competitive host-guest interactions in a turn-on fluorometric method to amantadine determination in human serum and pharmaceutical formulations

A competitive assay between the antiviral Amantadine and the dye Thionine for the Cucurbit[8]uril cavity was carried out in a flow injection analysis system for the indirect fluorescence detection of Amantadine. Both, Cucurbit[7]uril and Cucurbit[8]uril Thionine complexes were evaluated for the competitive assay. The use of a 12-port injection valve allows the on-line reaction in the flow system. Once optimized all the experimental variables, the methodology developed allows the detection of Amantadine at the 0.16µM level with excellent accuracy (Er ≤ 8.2%) and reproducibility (RSD ≤ 6.3%) for all the concentration range assayed. This one-step turn-on fluorescence methodology allows reaching sampling frequencies of 68 samples per hour. The selectivity of the method was evaluated against different antiviral drugs. Moreover, the performance of the methodology proposed was tested by the Amantadine determination in human serum and pharmaceutical formulations samples. The results demonstrated that the method can be applied to Amantadine determination in real samples of different nature with excellent recoveries, ranging from 83% to 98% depending on the matrix assayed.

Diffusion methods for wind power ramp detection

The prediction and management of wind power ramps is currently receiving large attention as it is a crucial issue for both system operators and wind farm managers. However, this is still an issue far from being solved and in this work we will address it as a classification problem working with delay vectors of the wind power time series and applying local Mahalanobis K-NN search with metrics derived from Anisotropic Diffusion methods. The resulting procedures clearly outperform a random baseline method and yield good sensitivity but more work is needed to improve on specificity and, hence, precision.

Diffusion maps for the description of meteorological data

Diffusion Maps is a new powerful technique for dimensionality reduction that can capture geometric structure while taking into account data distribution. In this work we will apply it to time and spatial compression of numerical weather forecasts, showing how it is capable to greatly reduce the initial dimension while still capturing relevant information in the original data.

Valor de la PET/TC cerebral con 18F-fluorocolina en la detección de recurrencias de neoplasias primarias del sistema nervioso central

AIM To study the usefulness of 18F-fluorocholine (FCH) in detecting the recurrence of primary brain tumours. MATERIAL AND METHODS A prospective study was conducted on brain PET/CT with FCH for compassionate use in 21 patients with suspected recurrence of a primary brain tumour. The distribution by pathology was: three grade ii astrocytomas, three grade iii astrocytomas, one grade ii oligodendroglioma, three grade iii oligodendrogliomas, one grade iii oligoastrocytoma, four glioblastoma multiform, one gliomatosis cerebri, and five meningiomas. Studies in which there was a visually significant uptake in the brain parenchyma were classified as positive. RESULTS A total of 17 patients were classified as positive, with the results being confirmed by histology (10 cases) or clinical follow-up and imaging, with no false positives or negatives. The mean SUVmax for positive patients was 8.02 and 0.94 for the negative ones, which was significantly different (P=.003) CONCLUSION: PET/CT with FCH shows encouraging results in the evaluation of patients with suspected recurrence of primary brain neoplasms.

Kernel K-Means Low Rank Approximation for Spectral Clustering and Diffusion Maps

Spectral Clustering and Diffusion Maps are currently the leading methods for advanced clustering or dimensionality reduction. However, they require the eigenanalysis of a sample’s graph Laplacian L, something very costly for moderately sized samples and prohibitive for very large ones. We propose to build a low rank approximation to L using essentially the centroids obtained applying kernel K-means over the similarity matrix. We call this approach kernel KASP (kKASP) as it follows the KASP procedure of Yan et al. but coupling centroid selection with the local geometry defined by the similarity matrix. As we shall see, kKASP’s reconstructions are competitive with KASP’s ones, particularly in the low rank range.