No description available.

This dissertation develops a comprehensive statistical learning framework that is robust to (distributional) perturbations in the data using Distributionally Robust Optimization (DRO) under the Wasserstein metric. The learning problems that are studied include: (i) Distributionally Robust Linear Regression (DRLR), which estimates a robustified linear regression plane by minimizing the worst-case expected absolute loss over a probabilistic ambiguity set characterized by the Wasserstein metric; (ii) Groupwise Wasserstein Grouped LASSO (GWGL), which aims at inducing sparsity at a group level when there exists a predefined grouping structure for the predictors, through defining a specially structured Wasserstein metric for DRO; (iii) Optimal decision making using DRLR informed K-Nearest Neighbors (K-NN) estimation, which selects among a set of actions the optimal one through predicting the outcome under each action using K-NN with a distance metric weighted by the DRLR solution; and (iv) Distributionally Robust Multivariate Learning, which solves a DRO problem with a multi-dimensional response/label vector, as in Multivariate Linear Regression (MLR) and Multiclass Logistic Regression (MLG), generalizing the univariate response model addressed in DRLR. A tractable DRO relaxation for each problem is being derived, establishing a connection between robustness and regularization, and obtaining upper bounds on the prediction and estimation errors of the solution. The accuracy and robustness of the estimator is verified through a series of synthetic and real data experiments. The experiments with real data are all associated with various health informatics applications, an application area which motivated the work in this dissertation. In addition to estimation (regression and classification), this dissertation also considers outlier detection applications.

The present work is about measure-valued diffusion processes, which are aligned with two distinct geometries on the set of probability measures. In the first part we focus on a stochastic partial differential equation, the Dean-Kawasaki equation, which can be considered as a natural candidate for a Langevin equation on probability measures, when equipped with the Wasserstein distance. Apart from that, the dynamic in question appears frequently as a model for fluctuating density fields in non-equilibrium statistical mechanics. Yet, we prove that the Dean-Kawasaki equation admits a solution only in integer parameter regimes, in which case the solution is given by a particle system of finite size with mean field interaction. For the second part we restrict ourselves to positive probability measures on a finite set, which we identify with the open standard unit simplex. We show that Brownian motion on the simplex equipped with the Aitchison geometry, can be interpreted as a replicator dynamic in a white noise fitness landscape. We infer three approximation results for this Aitchison diffusion. Finally, invoking Fokker-Planck equations and Wasserstein contraction estimates, we study the long time behavior of the stochastic replicator equation, as an example of a non-gradient drift diffusion on the Aitchison simplex.

[pt] A programação estocástica dois estágios é uma abordagem matemática amplamente usada em aplicações da vida real, como planejamento da operação de sistemas de energia, cadeias de suprimentos, logística, gerenciamento de inventário e planejamento financeiro. Como a maior parte desses problemas não pode ser resolvida analiticamente, os tomadores de decisão utilizam métodos numéricos para obter uma solução quase ótima. Em algumas aplicações, soluções não convergidas e, portanto, sub-ótimas terminam sendo implementadas devido a limitações de tempo ou esforço computacional. Nesse contexto, os métodos existentes fornecem uma solução otimista sempre que a convergência não é atingida. As soluções otimistas geralmente geram altos níveis de arrependimento porque subestimam os custos reais na função objetivo aproximada. Para resolver esse problema, temos desenvolvido duas metodologias de solução conservadora para problemas de programação linear estocástica dois estágios com incerteza do lado direito e suporte retangular: Quando a verdadeira distribuição de probabilidade da incerteza é conhecida, propomos um problema DRO (Distributionally Robust Optimization) baseado em esperanças condicionais adaptadas à uma partição do suporte cuja complexidade cresce exponencialmente com a dimensionalidade da incerteza; Quando apenas observações históricas da incerteza estão disponíveis, propomos um problema de DRO baseado na métrica de Wasserstein a fim de incorporar ambiguidade sobre a real distribuição de probabilidade da incerteza. Para esta última abordagem, os métodos existentes dependem da enumeração dos vértices duais do problema de segundo estágio, tornando o problema DRO intratável em aplicações práticas. Nesse contexto, propomos esquemas algorítmicos para lidar com a complexidade computacional de ambas abordagens. Experimentos computacionais são apresentados para o problema do fazendeiro, o problema de alocação de aviões, e o problema do planejamento da operação do sistema elétrico (unit ommitmnet problem). / [en] Two-stage stochastic programming is a mathematical framework widely used in real-life applications such as power system operation planning, supply chains, logistics, inventory management, and financial planning. Since most of these problems cannot be solved analytically, decision-makers make use of numerical methods to obtain a near-optimal solution. Some applications rely on the implementation of non-converged and therefore sub-optimal solutions because of computational time or power limitations. In this context, the existing methods provide an optimistic solution whenever convergence is not attained. Optimistic solutions often generate high disappointment levels because they consistently underestimate the actual costs in the approximate objective function. To address this issue, we have developed two conservative-solution methodologies for two-stage stochastic linear programming problems with right-hand-side uncertainty and rectangular support: When the actual data-generating probability distribution is known, we propose a DRO problem based on partition-adapted conditional expectations whose complexity grows exponentially with the uncertainty dimensionality; When only historical observations of the uncertainty are available, we propose a DRO problem based on the Wasserstein metric to incorporate ambiguity over the actual data-generating probability distribution. For this latter approach, existing methods rely on dual vertex enumeration of the second-stage problem rendering the DRO problem intractable in practical applications. In this context, we propose algorithmic schemes to address the computational complexity of both approaches. Computational experiments are presented for the farmer problem, aircraft allocation problem, and the stochastic unit commitment problem.

[pt] Uma medida de probabilidade com suporte compacto em um grupo de matrizes determina uma sequência de matrizes aleatórias i.i.d. Considere o processo multiplicativo correspondente e suas médias geométricas. O teorema de Furstenberg-Kesten, análogo da lei dos grandes números neste cenário, garante que as médias geométricas desse processo multiplicativo convergem quase certamente para uma constante, chamada de expoente de Lyapunov maximal da medida dada. Este conceito pode ser reformulado no contexto mais geral da teoria ergódica usando cociclos lineares aleatórios sobre o shift de Bernoulli. Uma questão natural diz respeito às propriedades de regularidade do expoente de Lyapunov como uma função dos seus dados. Sob uma condição de irredutibilidade e em um cenário específico (que foi posteriormente generalizado por vários autores) Le Page estabeleceu a continuidade de Holder do expoente de Lyapunov. Recentemente, Baraviera e Duarte obtiveram uma prova direta e elegante deste tipo de resultado. Seu argumento usa a fórmula de Furstenberg e as propriedades de regularidade da medida estacionária. Seguindo sua abordagem, neste trabalho obtemos um novo resultado mostrando que, sob a mesma hipótese de irredutibilidade, o expoente de Lyapunov depende Hölder continuamente da medida, relativamente à métrica de Wasserstein, generalizando assim o resultado de Baraviera e Duarte. / [en] A compactly supported probability measure on a group of matrices determines a sequence of i.i.d. random matrices. Consider the corresponding multiplicative process and its geometric averages. Furstenberg-Kesten s theorem, the analogue of the law of large numbers in this setting, ensures that the geometric averages of this multiplicative process converge almost surely to a constant, called the maximal Lyapunov exponent of the given measure. This concept can be reformulated in the more general context of ergodic theory using random linear cocycles over the Bernoulli shift. A natural question concerns the regularity properties of the Lyapunov exponent as a function of the data. Under an irreducibility condition and in a specific setting (which was later generalized by various authors) Le Page established the Holder continuity of the Lyapunov exponent. Recently, Baraviera and Duarte obtained a direct and elegant proof of this type of result. Their argument uses Furstenberg s formula and the regularity properties of the stationary measure. Following their approach, in this work we obtain a new result showing that under the same irreducibility hypothesis, the Lyapunov exponent depends Holder continuously on the measure, relative to the Wasserstein metric, thus generalizing the result of Baraviera and Duarte.