SEQUENTIAL DECISIONS AND PREDICTIONS IN NATURAL LANGUAGE PROCESSING

Natural language processing has achieved great success in a wide range of ap- plications, producing both commercial language services and open-source language tools. However, most methods take a static or batch approach, assuming that the model has all information it needs and makes a one-time prediction. In this disser- tation, we study dynamic problems where the input comes in a sequence instead of all at once, and the output must be produced while the input is arriving. In these problems, predictions are often made based only on partial information. We see this dynamic setting in many real-time, interactive applications. These problems usually involve a trade-off between the amount of input received (cost) and the quality of the output prediction (accuracy). Therefore, the evaluation considers both objectives (e.g., plotting a Pareto curve). Our goal is to develop a formal understanding of sequential prediction and decision-making problems in natural language processing and to propose efficient solutions. Toward this end, we present meta-algorithms that take an existent batch model and produce a dynamic model to handle sequential inputs and outputs. Webuild our framework upon theories of Markov Decision Process (MDP), which allows learning to trade off competing objectives in a principled way. The main machine learning techniques we use are from imitation learning and reinforcement learning, and we advance current techniques to tackle problems arising in our settings. We evaluate our algorithm on a variety of applications, including dependency parsing, machine translation, and question answering. We show that our approach achieves a better cost-accuracy trade-off than the batch approach and heuristic-based decision- making approaches. We first propose a general framework for cost-sensitive prediction, where dif- ferent parts of the input come at different costs. We formulate a decision-making process that selects pieces of the input sequentially, and the selection is adaptive to each instance. Our approach is evaluated on both standard classification tasks and a structured prediction task (dependency parsing). We show that it achieves similar prediction quality to methods that use all input, while inducing a much smaller cost. Next, we extend the framework to problems where the input is revealed incremen- tally in a fixed order. We study two applications: simultaneous machine translation and quiz bowl (incremental text classification). We discuss challenges in this set- ting and show that adding domain knowledge eases the decision-making problem. A central theme throughout the chapters is an MDP formulation of a challenging problem with sequential input/output and trade-off decisions, accompanied by a learning algorithm that solves the MDP.

EXISTENCE OF UNUSED MANAGED CAPACITY ON DEDICATED LANES AND AN ALTERNATIVE ON HOW TO SELL IT VIA AUCTIONS

This dissertation verifies whether the following two hypotheses are true: (1) High-occupancy/toll lanes (and therefore other dedicated lanes) have capacity that could still be used; (2) such unused capacity (or more precisely, “unused managed capacity”) can be sold successfully through a real-time auction. To show that the second statement is true, this dissertation proposes an auction-based metering (ABM) system, that is, a mechanism that regulates traffic that enters the dedicated lanes. Participation in the auction is voluntary and can be skipped by paying the toll or by not registering to the new system. This dissertation comprises the following four components: a measurement of unused managed capacity on an existing HOT facility, a game-theoretic model of an ABM system, an operational description of the ABM system, and a simulation-based evaluation of the system. Some other and more specific contributions of this dissertation include the following: (1) It provides a definition and a methodology for measuring unused managed capacity and another important variable referred as “potential volume increase”. (2) It proves that the game-theoretic model has a unique Bayesian Nash equilibrium. (3) And it provides a specific road design that can be applied or extended to other facilities. The results provide evidence that the hypotheses are true and suggest that the ABM system would benefit a public operator interested in reducing traffic congestion significantly, would benefit drivers when making low-reliability trips (such as work-to-home trips), and would potentially benefit a private operator interested in raising revenue.