IRGAN: A Minimax Game for Unifying Generative and Discriminative Information Retrieval Models (2017-05-30T00:00:00.000000Z)

TL;DR

A unified framework takes advantage of both schools of thinking in information retrieval modelling and shows that the generative model learns to fit the relevance distribution over documents via the signals from the discriminative model to achieve a better estimation for document ranking.

Abstract

This paper provides a unified account of two schools of thinking in information retrieval modelling: the generative retrieval focusing on predicting relevant documents given a query, and the discriminative retrieval focusing on predicting relevancy given a query-document pair. We propose a game theoretical minimax game to iteratively optimise both models. On one hand, the discriminative model, aiming to mine signals from labelled and unlabelled data, provides guidance to train the generative model towards fitting the underlying relevance distribution over documents given the query. On the other hand, the generative model, acting as an attacker to the current discriminative model, generates difficult examples for the discriminative model in an adversarial way by minimising its discrimination objective. With the competition between these two models, we show that the unified framework takes advantage of both schools of thinking: (i) the generative model learns to fit the relevance distribution over documents via the signals from the discriminative model, and (ii) the discriminative model is able to exploit the unlabelled data selected by the generative model to achieve a better estimation for document ranking. Our experimental results have demonstrated significant performance gains as much as 23.96% on Precision@5 and 15.50% on MAP over strong baselines in a variety of applications including web search, item recommendation, and question answering.

TL;DR

Abstract

Authors

References55 items

A Distribution Separation Method Using Irrelevance Feedback Data for Information Retrieval

NIPS 2016 Tutorial: Generative Adversarial Networks

Dual Learning for Machine Translation

LambdaFM: Learning Optimal Ranking with Factorization Machines Using Lambda Surrogates

SeqGAN: Sequence Generative Adversarial Nets with Policy Gradient

Deep Neural Networks for YouTube Recommendations

Improved Techniques for Training GANs

ITEM2VEC: Neural item embedding for collaborative filtering

Attentive Pooling Networks

How (not) to Train your Generative Model: Scheduled Sampling, Likelihood, Adversary?

Learning to Rank Short Text Pairs with Convolutional Deep Neural Networks

Applying deep learning to answer selection: A study and an open task

A Long Short-Term Memory Model for Answer Sentence Selection in Question Answering

On distinguishability criteria for estimating generative models

Conditional Generative Adversarial Nets

Convolutional Neural Networks for Sentence Classification

Win-win search: dual-agent stochastic game in session search

Generative Adversarial Nets

Improving pairwise learning for item recommendation from implicit feedback

Distributed Representations of Words and Phrases and their Compositionality

Ad click prediction: a view from the trenches

Optimizing top-n collaborative filtering via dynamic negative item sampling

Factorization Machines

From RankNet to LambdaRank to LambdaMART: An Overview

Noise-contrastive estimation: A new estimation principle for unnormalized statistical models

Ranking Measures and Loss Functions in Learning to Rank

Matrix Factorization Techniques for Recommender Systems

BPR: Bayesian Personalized Ranking from Implicit Feedback

McRank: Learning to Rank Using Multiple Classification and Gradient Boosting

Learning to rank: from pairwise approach to listwise approach

Learning to Rank with Nonsmooth Cost Functions

Unifying user-based and item-based collaborative filtering approaches by similarity fusion

Regularized estimation of mixture models for robust pseudo-relevance feedback

Learning to rank using gradient descent

Discriminative models for information retrieval

A study of smoothing methods for language models applied to information retrieval

Improving pseudo-relevance feedback in web information retrieval using web page segmentation

Tutorial on maximum likelihood estimation

Optimizing search engines using clickthrough data

Model-based feedback in the language modeling approach to information retrieval

Policy Gradient Methods for Reinforcement Learning with Function Approximation

Policy Invariance Under Reward Transformations: Theory and Application to Reward Shaping

A Language Modeling Approach to Information Retrieval

An Efficient Boosting Algorithm for Combining Preferences

Simple statistical gradient-following algorithms for connectionist reinforcement learning

Relevance weighting of search terms

Towards a game-theoretic framework for text data retrieval

LETOR: Benchmark Dataset for Research on Learning to Rank for Information Retrieval

Simple Statistical Gradient-Following Algorithms for Connectionist Reinforcement Learning

Probabilistic Relevance Models Based on Document and Query Generation

An EÆcient Boosting Algorithm for Combining Preferences

Large Margin Rank Boundaries for Ordinal Regression

Large margin rank boundaries for ordinal regression

Probabilistic latent semantic indexing

A language modeling approach to information retrieval

Field of Study

Journal Information

Name

Venue Information

Name

Type

URL

Alternate Names