Query minimization under stochastic uncertainty

2020

We study how to utilize (possibly machine-learned) predictions in a model for optimization under uncertainty that allows an algorithm to query unknown data. The goal is to minimize the number of queries needed to solve the problem. Considering fundamental problems such as finding the minima of intersecting sets of elements or sorting them, as well as the minimum spanning tree problem, we discuss different measures for the prediction accuracy and design algorithms with performance guarantees that improve with the accuracy of predictions and that are robust with respect to very poor prediction quality. We also provide new structural insights for the minimum spanning tree problem that might be useful in the context of explorable uncertainty regardless of predictions. Our results prove that untrusted predictions can circumvent known lower bounds in the model of explorable uncertainty. We complement our results by experiments that empirically confirm the performance of our algorithms.

arXiv (Cornell University), 2022

We study how to utilize (possibly erroneous) predictions in a model for computing under uncertainty in which an algorithm can query unknown data. Our aim is to minimize the number of queries needed to solve the minimum spanning tree problem, a fundamental combinatorial optimization problem that has been central also to the research area of explorable uncertainty. For all integral γ ≥ 2, we present algorithms that are γ-robust and (1+ 1 γ)-consistent, meaning that they use at most γOPT queries if the predictions are arbitrarily wrong and at most (1+ 1 γ)OPT queries if the predictions are correct, where OPT is the optimal number of queries for the given instance. Moreover, we show that this trade-off is best possible. Furthermore, we argue that a suitably defined hop distance is a useful measure for the amount of prediction error and design algorithms with performance guarantees that degrade smoothly with the hop distance. We also show that the predictions are PAC-learnable in our model. Our results demonstrate that untrusted predictions can circumvent the known lower bound of 2, without any degradation of the worst-case ratio. To obtain our results, we provide new structural insights for the minimum spanning tree problem that might be useful in the context of query-based algorithms regardless of predictions. In particular, we generalize the concept of witness sets-the key to lower-bounding the optimum-by proposing novel global witness set structures and completely new ways of adaptively using those. 2012 ACM Subject Classification Theory of Computation → Design and analysis of algorithms Keywords and phrases explorable uncertainty, queries, untrusted predictions

2019

We study the problem of sorting under incomplete information, when queries are used to resolve uncertainties. Each of n data items has an unknown value, which is known to lie in a given interval. We can pay a query cost to learn the actual value, and we may allow an error threshold in the sorting. The goal is to find a nearly-sorted permutation by performing a minimum-cost set of queries. We show that an offline optimum query set can be found in polynomial time, and that both oblivious and adaptive problems have simple query-competitive algorithms. The query-competitiveness for the oblivious problem is n for uniform query costs, and unbounded for arbitrary costs; for the adaptive problem, the ratio is 2. We then present a unified adaptive strategy for uniform query costs that yields: (i) a 3/2-query-competitive randomized algorithm; (ii) a 5/3-query-competitive deterministic algorithm if the dependency graph has no 2-components after some preprocessing, which has query-competitive r...

arXiv (Cornell University), 2020

We study how to utilize (possibly machine-learned) predictions in a model for computing under uncertainty in which an algorithm can query unknown data. The goal is to minimize the number of queries needed to solve the problem. We consider fundamental problems such as finding the minima of intersecting sets of elements or sorting them (these problems can also be phrased as (hyper)graph orientation problems), as well as the minimum spanning tree problem. We discuss different measures for the prediction accuracy and design algorithms with performance guarantees that improve with the accuracy of predictions and that are robust with respect to very poor prediction quality. These measures are intuitive and might be of general interest for inputs involving uncertainty intervals. We show that our predictions are PAC learnable. We also provide new structural insights for the minimum spanning tree problem that might be useful in the context of explorable uncertainty regardless of predictions. Our results prove that untrusted predictions can circumvent known lower bounds in the model of explorable uncertainty. We complement our results by experiments that empirically confirm the performance of our algorithms.

Most work on sequential learning assumes a fixed set of actions that are available all the time. However, in practice, actions can consist of picking subsets of readings from sensors that may break from time to time, road segments that can be blocked or goods that are out of stock. In this paper we study learning algorithms that are able to deal with stochastic availability of such unreliable composite actions. We propose and analyze algorithms based on the Follow-The-Perturbed-Leader prediction method for several learning settings differing in the feedback provided to the learner. Our algorithms rely on a novel loss estimation technique that we call Counting Asleep Times. We deliver regret bounds for our algorithms for the previously studied full information and (semi-)bandit settings, as well as a natural middle point between the two that we call the restricted information setting. A special consequence of our results is a significant improvement of the best known performance guarantees achieved by an efficient algorithm for the sleeping bandit problem with stochastic availability. Finally, we evaluate our algorithms empirically and show their improvement over the known approaches.

Discrete Applied Mathematics, 1999

Kearns introduced the "statistical query" (SQ) model as a general method for producing learning algorithms which are robust against classification noise. We extend this approach in several ways in order to tackle algorithms that use "membership queries", focusing on the more stringent model of "persistent noise". The main ingredients in the general analysis are:

Proceedings of the AAAI Conference on Artificial Intelligence

This paper studies bandit algorithms under data poisoning attacks in a bounded reward setting. We consider a strong attacker model in which the attacker can observe both the selected actions and their corresponding rewards, and can contaminate the rewards with additive noise. We show that any bandit algorithm with regret O(log T) can be forced to suffer a regret O(T) with an expected amount of contamination O(log T). This amount of contamination is also necessary, as we prove that there exists an O(log T) regret bandit algorithm, specifically the classical UCB, that requires Omega(log T) amount of contamination to suffer regret Omega(T). To combat such poisoning attacks, our second main contribution is to propose verification based mechanisms, which use limited verification to access a limited number of uncontaminated rewards. In particular, for the case of unlimited verifications, we show that with O(log T) expected number of verifications, a simple modified version of the Explore-...

Information Processing Letters, 1998

We consider the problem of identifying an unknown value x ∈ {1, 2,. .. , n} by asking "Yes-No" questions about x. The goal is to minimize the number of questions required in the worst case, taking into account that no more than B questions may receive answer "Yes" and no more than E answers may be erroneous. We consider two versions of this problem: the discrete and the continuous version. In the discrete case x is a member of the finite set {1, 2,. .. , n}; in the continuous case x is a member of the half-open real interval (0, 1]. In the continuous case we will not in general be able to identify x exactly with a finite number of questions; rather we fix a size ε and then we compute the exact value of the minimal number of questions to get a subset of (0, 1] of size ε which is known to contain x. The solution of the continuous version allows us to derive a lower bound for the minimal number of questions required for the discrete version of the problem.

Proceedings of the 11th international conference on Extending database technology Advances in database technology - EDBT '08, 2008

Recently, many new applications, such as sensor data monitoring and mobile device tracking, raise up the issue of uncertain data management. Compared to "certain" data, the data in the uncertain database are not exact points, which, instead, often locate within a region. In this paper, we study the ranked queries over uncertain data. In fact, ranked queries have been studied extensively in traditional database literature due to their popularity in many applications, such as decision making, recommendation raising, and data mining tasks. Many proposals have been made in order to improve the efficiency in answering ranked queries. However, the existing approaches are all based on the assumption that the underlying data are exact (or certain). Due to the intrinsic differences between uncertain and certain data, these methods are designed only for ranked queries in certain databases and cannot be applied to uncertain case directly. Motivated by this, we propose novel solutions to speed up the probabilistic ranked query (PRank) over the uncertain database. Specifically, we introduce two effective pruning methods, spatial and probabilistic, to help reduce the PRank search space. Then, we seamlessly integrate these pruning heuristics into the PRank query procedure. Extensive experiments have demonstrated the efficiency and effectiveness of our proposed approach in answering PRank queries, in terms of both wall clock time and the number of candidates to be refined.

Proceedings of the thirty-second annual ACM symposium on Theory of computing - STOC '00, 2000

We consider a class of problems in which an algorithm seeks to compute a function f over a set of n inputs, where each input has an associated price. The algorithm queries inputs sequentially, trying to learn the value of the function for the minimum cost. We apply the competitive analysis of algorithms to this framework, designing algorithms that incur large cost only when the cost of the cheapest "proof" for the value of f is also large. We provide algorithms that achieve the optimal competitive ratio for functions that include arbitrary Boolean AND/OR trees, and for the problem of searching in a sorted array. We also investigate a model for pricing in this framework, constructing a set of prices for any AND/OR tree that satisfies a very strong type of equilibrium property.