Theses and Dissertations from UMD
Permanent URI for this communityhttp://hdl.handle.net/1903/2
New submissions to the thesis/dissertation collections are added automatically as they are received from the Graduate School. Currently, the Graduate School deposits all theses and dissertations from a given semester after the official graduation date. This means that there may be up to a 4 month delay in the appearance of a give thesis/dissertation in DRUM
More information is available at Theses and Dissertations at University of Maryland Libraries.
Browse
Search Results
Item Online Decision Making via Prophet Setting(2017) Ehsani Banafati, Soheil; Hajiaghayi, MohammadTaghi; Computer Science; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)In the study of online problems, it is often assumed that there exists an adversary who acts against the algorithm and generates the most challenging input for it. This worst-case assumption in addition to the complete uncertainty about future events in the traditional online setting sometimes leads to worst-case scenarios with super-constant approximation impossibilities. In this dissertation, we go beyond this worst-case analysis of problems by taking advantage of stochastic modeling. Inspired by the prophet inequality problem, we introduce the prophet setting for online problems in which the probability distributions of the future inputs are available. This modeling not only considers the availability of statistical data in the design of mechanisms but also results in significantly more efficient algorithms. To illustrate the improvements achieved by this setting, we study online problems within the contexts of auctions and networks. We begin our study with analyzing a fundamental online problem in optimal stopping theory, namely prophet inequality, in the special cases of iid and large markets, and general cases of matroids and combinatorial auctions and discuss its applications in mechanism design. The stochastic model introduced by this problem has received a lot of attention recently in modeling other real-life scenarios, such as online advertisement, because of the growing ability to fit distributions for user demands. We apply this model to network design problems with a wide range of applications from social networks to power grids and communication networks. In this dissertation, we give efficient algorithms for fundamental network design problems in the prophet setting and present a general framework that demonstrates how to develop algorithms for other problems in this setting.Item ALLOCATIONS IN LARGE MARKETS(2017) Esfandiari, Hossein; Hajiaghayi, MohammadTaghi; Computer Science; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Rapid growth and popularity of internet based services such as online markets and online advertisement systems provide a lot of new algorithmic challenges. One of the main challenges is the limited access to the input. There are two main reasons that algorithms have limited data accessibility. 1) The input is extremely large, and hence having access to the whole data at once is not practical. 2) The nature of the system forces us to make decisions before observing the whole input. Internet-enabled marketplaces such as Amazon and eBay deal with huge datasets registering transaction of merchandises between lots of buyers and sellers. It is important that algorithms become more time and space efficient as the size of datasets increase. An algorithm that runs in polynomial time may not have a reasonable running time for such large datasets. In the first part of this dissertation, we study the development of allocation algorithms that are appropriate for use with massive datasets. We especially focus on the streaming setting which is a common model for big data analysis. In the graph streaming, the algorithm has access to a sequence of edges, called a stream. The algorithm reads edges in the order in which they appear in the stream. The goal is to design an algorithm that maintains a large allocation, using as little space as possible. We achieve our results by developing powerful sampling techniques. Indeed, one can implement our sampling techniques in the streaming setting as well as other distributed settings such as MapReduce. Giant online advertisement markets such as Google, Bing and Facebook raised up several interesting allocation problems. Usually, in these applications, we need to make the decision before obtaining the full information of the input graph. This enforces an uncertainty on our belief about the input, and thus makes the classical algorithms inapplicable. To address this shortcoming online algorithms have been developed. In online algorithms again the input is a sequence of items. Here the algorithm needs to make an irrevocable decision upon arrival of each item. In the second part of this dissertation, we aim to achieve two main goals for each allocation problem in the market. Our first goal is to design models to capture the uncertainty of the input based on the properties of problems and the accessible data in real applications. Our second goal is to design algorithms and develop new techniques for these market allocation problems.