Analysis of Watson's Strategies for Playing Jeopardy!

  title={Analysis of Watson's Strategies for Playing Jeopardy!},
  author={Gerald Tesauro and David Gondek and Jonathan Lenchner and James Fan and John M. Prager},
  journal={J. Artif. Intell. Res.},
Major advances in Question Answering technology were needed for IBM Watson1 to play Jeopardy! at championship level - the show requires rapid-fire answers to challenging natural language questions, broad general knowledge, high precision, and accurate confidence estimates. In addition, Jeopardy! features four types of decision making carrying great strategic importance: (1) Daily Double wagering; (2) Final Jeopardy wagering; (3) selecting the next square when in control of the board; (4… 

The Psychology of Big Data: Developing a 'Theory of Machine' to Examine Perceptions of Algorithms

Organizations cannot realize the full potential of algorithms until they address the last mile problem and consider how people respond to algorithmic advice, as most overlook the important connection between producing and utilizing insights.

Quizbowl: The Case for Incremental Question Answering

This work makes two key contributions to machine learning research through Quizbowl: collecting and curating a large factoid QA dataset and an accompanying gameplay dataset, and developing a computational approach to playing Quiz Bowl that involves determining both what to answer and when to answer.

When do people rely on algorithms

In eight experiments, this work tested whether aversion to algorithms is as straightforward a story as past work suggests and shed light on the important questions of when people rely on algorithmic advice over advice from people and have implications for the use of algorithms within organizations.

The Persuasive Power of Algorithmic and Crowdsourced Advice

Examining the effects of algorithmic and social advice on decision-making in the context of an online retirement saving system finds that both types of advice have a positive effect on users’ saving performance, and that users follow advice presented as coming from an algorithmic source more closely than Advice presented as crowdsourced.

Imagination Machines: A New Challenge for Artificial Intelligence

It is argued for the development of a new field called imagination science, which extends data science beyond its current realm of learning probability distributions from samples, and why automation of imagination provides a powerful launching pad for transforming AI.

Design of Artificial Intelligence Agents for Games using Deep Reinforcement Learning

The results of this investigation open the way into new, biologically-inspired deep reinforcement learning algorithms, and their implementation on neuromorphic hardware.

The Eigenoption-Critic Framework

An algorithm termed eigenoption-critic (EOC) based on the Option-Critic (OC) framework is introduced, a general hierarchical reinforcement learning (RL) algorithm that allows learning the intra-option policies simultaneously with the policy over options.

Wanted: Collaborative intelligence

Knowledge Extraction in Web Media: At The Frontier of NLP, Machine Learning and Semantics

This research presents a preliminary framework based on a novel hybrid architecture for an entity linking system, that combines methods from the Natural Language Processing (NLP), information retrieval and semantic fields, and proposes a modular approach in order to be as independent as possible of the text to be processed.



Simulation, learning, and optimization techniques in Watson's game strategies

Application of machine learning and Monte Carlo methods used in simulations to optimize the respective strategy algorithms yielded superhuman game strategies for IBM Watsoni that significantly enhanced its overall competitive record.

Introduction to "This is Watson"

A brief history of the events and ideas that positioned the team to take on the Jeopardy! challenge, build Watson, IBM Watson™, and ultimately triumph is provided, and how the system performed at champion levels is summarized.

Special Questions and techniques

The design of the Special Question solving procedures motivated architectural design decisions that are applicable to general open-domain question-answering systems and are an important class of question to address in the Jeopardy! context.

Building Watson: An Overview of the DeepQA Project

The results strongly suggest that DeepQA is an effective and extensible architecture that may be used as a foundation for combining, deploying, evaluating and advancing a wide range of algorithmic techniques to rapidly advance the field of QA.

On-line Policy Improvement using Monte-Carlo Search

A Monte-Carlo simulation algorithm for real-time policy improvement of an adaptive controller and results are reported for a wide variety of initial policies, ranging from a random policy to TD-Gammon, an extremely strong multi-layer neural network.

Reinforcement Learning: An Introduction

This book provides a clear and simple account of the key ideas and algorithms of reinforcement learning, which ranges from the history of the field's intellectual foundations to the most recent developments and applications.

GIB: Steps Toward an Expert-Level Bridge-Playing Program

GIB, the first bridge-playing program to approach the level of a human expert, is described and the results of experiments comparing GIB to both human opponents and other programs are presented.

World-championship-caliber Scrabble

Game theory.

  • M. Dufwenberg
  • Psychology
    Wiley interdisciplinary reviews. Cognitive science
  • 2011
The nature of game-theoretic analysis, the history of game theory,Why game theory is useful for understanding human psychology, and why game theory has played a key role in the recent explosion of interest in the field of behavioral economics are discussed.

On the generation of correlated artificial binary data

This paper presents a computationally fast method to simulate multivariate binary distributions with a given correlation structure, and main interest is in the segmentation of marketing data, where data come from customer questionnaires with "yes/no" questions.