Optimal Approximation for Submodular and Supermodular Optimization with Bounded Curvature

We design new approximation algorithms for the problems of optimizing submodular and supermodular functions subject to a single matroid constraint. Specifically, we consider the case in which we wish to maximize a monotone increasing submodular function or minimize a monotone decreasing supermodular function with a bounded total curvature c. Intuitively, the parameter c represents how nonlinear a function f is: when c = 0, f is linear, while for c = 1, f may be an arbitrary monotone increasing submodular function. For the case of submodular maximization with total curvature c, we obtain a (1 − c/e)-approximation—the first improvement over the greedy algorithm of of Conforti and Cornuéjols from 1984, which holds for a cardinality constraint, as well as a recent analogous result for an arbitrary matroid constraint. Our approach is based on modifications of the continuous greedy algorithm and nonoblivious local search, and allows us to approximately maximize the sum of a nonnegative, monotone increasing submodular function and a (possibly negative) linear function. We show how to reduce both submodular maximization and supermodular minimization to this general problem when the objective function has bounded total curvature. We prove that the approximation results we obtain are the best possible in the value oracle model, even in the case of a cardinality constraint. We define an extension of the notion of curvature to general monotone set functions and show a (1 − c)-approximation for maximization and a 1/(1 − c)-approximation for minimization cases. Finally, we give two concrete applications of our results in the settings of maximum entropy sampling, and the column-subset selection problem.

Robust Adaptive Decision Making: Bayesian Optimization and Beyond

Ilija Bogunovic

The central task in many interactive machine learning systems can be formalized as the sequential optimization of a black-box function. Bayesian optimization (BO) is a powerful model-based framework for \emph{adaptive} experimentation, where the primary goal is the optimization of the black-box function via sequentially chosen decisions. In many real-world tasks, it is essential for the decisions to be \emph{robust} against, e.g., adversarial failures and perturbations, dynamic and time-varying phenomena, a mismatch between simulations and reality, etc. Under such requirements, the standard methods and BO algorithms become inadequate. In this dissertation, we consider four research directions with the goal of enhancing robust and adaptive decision making in BO and associated problems. First, we study the related problem of level-set estimation (LSE) with Gaussian Processes (GPs). While in BO the goal is to find a maximizer of the unknown function, in LSE one seeks to find all "sufficiently good" solutions. We propose an efficient confidence-bound based algorithm that treats BO and LSE in a unified fashion. It is effective in settings that are non-trivial to incorporate into existing algorithms, including cases with pointwise costs, heteroscedastic noise, and multi-fidelity setting. Our main result is a general regret guarantee that covers these aspects. Next, we consider GP optimization with robustness requirement: An adversary may perturb the returned design, and so we seek to find a robust maximizer in the case this occurs. This requirement is motivated by, e.g., settings where the functions during optimization and implementation stages are different. We propose a novel robust confidence-bound based algorithm. The rigorous regret guarantees for this algorithm are established and complemented with an algorithm-independent lower bound. We experimentally demonstrate that our robust approach consistently succeeds in finding a robust maximizer while standard BO methods fail. We then investigate the problem of GP optimization in which the reward function varies with time. The setting is motivated by many practical applications in which the function to be optimized is not static. We model the unknown reward function via a GP whose evolution obeys a simple Markov model. Two confidence-bound based algorithms with the ability to "forget" about old data are proposed. We obtain regret bounds for these algorithms that jointly depend on the time horizon and the rate at which the function varies. Finally, we consider the maximization of a set function subject to a cardinality constraint

k

in the case a number of items

\tau

from the returned set may be removed. One notable application is in batch BO where we need to select experiments to run, but some of them can fail. Our focus is on the worst-case adversarial setting, and we consider both \emph{submodular} (i.e., satisfies a natural notion of diminishing returns) and \emph{non-submodular} objectives. We propose robust algorithms that achieve constant-factor approximation guarantees. In the submodular case, the result on the maximum number of allowed removals is improved to

\tau = o(k)

in comparison to the previously known

\tau=o(\sqrt{k})

. In the non-submodular case, we obtain new guarantees in the support selection and batch BO tasks. We empirically demonstrate the robust performance of our algorithms in these, as well as, in data summarization and influence maximization tasks.

EPFL2019

How to Match in Modern Computational Settings

Buddhima Ruwanmini Gamlath Gamlath Ralalage

This thesis focuses on the maximum matching problem in modern computational settings where the algorithms have to make decisions with partial information.First, we consider two stochastic models called query-commit and price-of-information where the algorithm only knows the distribution from which the edges are sampled.In the query-commit model, the algorithm must query edges to know if they exist and is committed to adding all queried edges that exist to its output.In the price-of-information model, the algorithm incurs costs for querying edges, and the total query cost is subtracted from the output matching's weight.For maximum weighted matching in these models, previously known best algorithms were greedy algorithms that achieve 1/2 approximations. We improve the approximation ratio to 1 - 1/e in both models. Next, we consider situations where the input graphs do not fit into the space available for an algorithm instance. We consider two such models: the semi-streaming model where the algorithm receives the input as a stream of edges and the algorithm has only sub-linear (in the number of edges) space, and the massively parallel computation (MPC) model where the input is distributed among several machines, each of which has sub-linear space, and algorithm instances running on different machines must communicate in synchronous rounds.We start with a particular case of the semi-streaming model where the edges arrive in uniformly random order, and the algorithm goes over the stream only once. For this setting, we give the first algorithm that finds a (1/2 + c)-approximate maximum weighted matching in expectation; such algorithms were previously known only for the unweighted graphs.We then show how to efficiently find (1 - epsilon)-approximate weighted matchings for any epsilon > 0 in multi-pass semi-streaming and MPC models by extending our algorithmic ideas used in the single-pass semi-streaming model with random order edge arrivals.Finally, we study online algorithms for matching, where the input graph is gradually revealed over time. In the online edge-arrival setting, the graph is revealed one edge at a time, and an algorithm is forced to make irrevocable decisions on whether to add each edge to the output matching upon their arrival. We show that no online algorithm can achieve a competitive ratio of 1/2 + c for any constant c > 0 in this setting.In the online vertex-arrival setting, the graph is revealed one vertex at a time, together with its incident edges to already revealed vertices, and the algorithm must irrevocably decide to ignore the revealed vertex or match it to one of the available neighbors.In this setting, we show how to round a previously known fractional online matching algorithm to get an integral online matching algorithm with a competitive ratio of 1/2 + c for some constant c > 0.

EPFL2021