Machine Learning
http://hdl.handle.net/20.500.11824/12
Sun, 20 Sep 2020 18:31:27 GMT2020-09-20T18:31:27ZApplication of machine learning techniques to weather forecasting
http://hdl.handle.net/20.500.11824/1141
Application of machine learning techniques to weather forecasting
Rozas Larraondo P.
Weather forecasting is, still today, a human based activity. Although computer simulations play a major role in modelling the state and evolution of the atmosphere,
there is a lack of methodologies to automate the interpretation of the information
generated by these models. This doctoral thesis explores the use of machine learning
methodologies to solve specific problems in meteorology and particularly focuses
on the exploration of methodologies to improve the accuracy of numerical weather
prediction models using machine learning. The work presented in this manuscript
contains two different approaches using machine learning. In the first part, classical
methodologies, such as multivariate non-parametric regression and binary trees are
explored to perform regression on meteorological data. In this first part, we particularly focus on forecasting wind, where the circular nature of this variable opens
interesting challenges for classic machine learning algorithms and techniques. The
second part of this thesis, explores the analysis of weather data as a generic structured prediction problem using deep neural networks. Neural networks, such as
convolutional and recurrent networks provide a method for capturing the spatial
and temporal structure inherent in weather prediction models. This part explores
the potential of deep convolutional neural networks in solving difficult problems in
meteorology, such as modelling precipitation from basic numerical model fields. The
research performed during the completion of this thesis demonstrates that collaboration between the machine learning and meteorology research communities is mutually beneficial and leads to advances in both disciplines. Weather forecasting models
and observational data represent unique examples of large (petabytes), structured
and high-quality data sets, that the machine learning community demands for developing the next generation of scalable algorithms.
Wed, 24 Oct 2018 00:00:00 GMThttp://hdl.handle.net/20.500.11824/11412018-10-24T00:00:00ZTheoretical and Methodological Advances in Semi-supervised Learning and the Class-Imbalance Problem
http://hdl.handle.net/20.500.11824/1140
Theoretical and Methodological Advances in Semi-supervised Learning and the Class-Imbalance Problem
Ortigosa-Hernandez J.
his paper focuses on the theoretical and practical generalization of two known and challenging situations from the field of machine learning to classification problems in which the assumption of having a single binary class is not fulfilled.semi-supervised learning is a technique that uses large amounts of unlabeled data to improve the performance of supervised learning when the labeled data set is very limited. Specifically, this work contributes with powerful and computationally efficient methodologies to learn, in a semi-supervised way, classifiers for multiple class variables. Also, the fundamental limits of semi-supervised learning in multi-class problems are investigated in a theoretical way. The problem of class unbalance appears when the target variables present a probability distribution unbalanced enough to distort the solutions proposed by the traditional supervised learning algorithms. In this project, a theoretical framework is proposed to separate the deviation produced by class unbalance from other factors that affect the accuracy of classifiers. This framework is mainly used to make a recommendation of classifier assessment metrics in this situation. Finally, a measure of the degree of class unbalance in a data set correlated with the loss of accuracy caused is also proposed.
Fri, 30 Nov 2018 00:00:00 GMThttp://hdl.handle.net/20.500.11824/11402018-11-30T00:00:00ZKernels of Mallows Models under the Hamming Distance for solving the Quadratic Assignment Problem
http://hdl.handle.net/20.500.11824/1138
Kernels of Mallows Models under the Hamming Distance for solving the Quadratic Assignment Problem
Arza E.; Pérez A.; Irurozki E.; Ceberio J.
The Quadratic Assignment Problem (QAP) is a well-known permutation-based combinatorial optimization problem with real applications in industrial and logistics environments. Motivated by the challenge that this NP-hard problem represents, it has captured the attention of the optimization community for decades. As a result, a large number of algorithms have been proposed to tackle this problem. Among these, exact methods are only able to solve instances of size $n<40$. To overcome this limitation, many metaheuristic methods have been applied to the QAP.
In this work, we follow this direction by approaching the QAP through Estimation of Distribution Algorithms (EDAs). Particularly, a non-parametric distance-based exponential probabilistic model is used. Based on the analysis of the characteristics of the QAP, and previous work in the area, we introduce Kernels of Mallows Model under the Hamming distance to the context of EDAs. Conducted experiments point out that the performance of the proposed algorithm in the QAP is superior to (i) the classical EDAs adapted to deal with the QAP, and also (ii) to the specific EDAs proposed in the literature to deal with permutation problems.
Wed, 01 Jul 2020 00:00:00 GMThttp://hdl.handle.net/20.500.11824/11382020-07-01T00:00:00ZAn adaptive neuroevolution-based hyperheuristic
http://hdl.handle.net/20.500.11824/1137
An adaptive neuroevolution-based hyperheuristic
Arza E.; Ceberio J.; Pérez A.; Irurozki E.
According to the No-Free-Lunch theorem, an algorithm that performs efficiently on any type of problem does not exist. In this
sense, algorithms that exploit problem-specific knowledge usually outperform more generic approaches, at the cost of a more complex design and parameter tuning process. Trying to combine the best of both worlds, the field of hyperheuristics investigates the automatized generation and hybridization of heuristic algorithms.
In this paper, we propose a neuroevolution-based hyperheuristic approach. Particularly, we develop a population-based hyperheuristic algorithm that first trains a neural network on an instance of a problem and then uses the trained neural network to control how and which low-level operators are applied to each of the solutions when optimizing different problem instances. The trained neural network maps the state of the optimization process to the operations to be applied to the solutions in the population at each generation.
Wed, 01 Jan 2020 00:00:00 GMThttp://hdl.handle.net/20.500.11824/11372020-01-01T00:00:00Z