No abstract

No abstract

Automated hyperparameter tuning aspires to facilitate the application of machine learning for non-experts. In the literature, different optimization approaches are applied for that purpose. This paper investigates the performance of Differential Evolution for tuning hyperparameters of supervised learning algorithms for classification tasks. This empirical study involves a range of different machine learning algorithms and datasets with various characteristics to compare the performance of Differential Evolution with Sequential Model-based Algorithm Configuration (SMAC), a reference Bayesian Optimization approach. The results indicate that Differential Evolution outperforms SMAC for most datasets when tuning a given machine learning algorithm -particularly when breaking ties in a firstto-report fashion. Only for the tightest of computational budgets SMAC performs better. On small datasets, Differential Evolution outperforms SMAC by 19% (37% after tie-breaking). In a second experiment across a range of representative datasets taken from the literature, Differential Evolution scores 15% (23% after tiebreaking) more wins than SMAC.

Surface solar radiation forecasting permits to predict photovoltaic plant production for a massive and safe integration of solar energy into the electric network. For short-term forecasts (intra-day), methods using images from meteorological geostationary satellites are more suitable than numerical weather prediction models. Forecast schemes consist in assessing cloud motion vectors and in extrapolating cloud patterns from a given satellite image in order to predict cloud cover state above a PV plant. Atmospheric motion vectors retrieval techniques have been studied for several decades in order to improve weather forecasts. However, solar energy forecasting requires the extraction of cloud motion vectors on a finer spatial-and time-resolution than those provided for weather forecast applications. Even if motion vector retrieval is a wide research field in image processing related topics, only block-matching techniques are operationally used for solar energy forecasts via satellite images. In this paper, we propose two motion vectors extraction methods originating from video compression techniques (correlation phase and optical flow methods). We implemented them on a 6-day dataset of Meteosat-10 satellite diurnal images. We proceeded to cloud pattern extrapolation and compared predicted cloud maps against actual ones at different time horizons from 15 minutes to 4 hours ahead. Forecast scores were compared to the state-of-the-art (block matching) method. Correlation phase methods do not outperform block-matching but their computation time is about 25 times shorter. Optical flow based method outperforms all the methods with a satisfactory time computing.

No abstract

No abstract

No abstract

No abstract