Active Learning in Recommender Systems
Published 5 October 2010
Neil Rubens, Dain Kaplan, Masashi Sugiyama
Citations181
Generate an AI Snapshot to get a quick, structured summary of this paper.
Study Snapshot
ObjectiveStudy objective
MethodsResearch methodology
PopulationPopulation studied
Sample sizeSample sizes
OutcomesStudy outcomes here
ResultsStudy results comes here
LimitationsResearch study limitations comes here
A concise AI-generated summary of the paper will appear here once you click Generate AI Snapshot.
TL;DR
Augmenting RSs with AL helps the user become more self-aware of their own likes/dislikes while at the same time providing new information to the system that it can analyze for subsequent recommendations.
Abstract
Recommender Systems (RSs) are often assumed to present items to users for one reason – to recommend items a user will likely be interested in. Of course RSs do recommend, but this assumption is biased, with no help of the title, towards the "recommending" the system will do. There is another reason for presenting an item to the user: to learn more about his/her preferences, or his/her likes and dislikes. This is where Active Learning (AL) comes in. Augmenting RSs with AL helps the user become more self-aware of their own likes/dislikes while at the same time providing new information to the system that it can analyze for subsequent recommendations. In essence, applying AL to RSs allows for personalization of the recommending process, a concept that makes sense as recommending is inherently geared towards personalization. This is accomplished by letting the system actively influence which items the user is exposed to (e.g. the items displayed to the user during sign-up orduring regular use), and letting the user explore his/her interests freely.
Keywords
Computer ScienceDecision Sciences
Journal of Computer and System SciencesA Decision-Theoretic Generalization of On-Line Learning and an Application to Boosting
20,320 Citations1997Yoav Freund, Robert E. Schapire
Machine LearningBagging predictors
16,377 Citations1996Leo Breiman
Tests on real and simulated data sets using classification and regression trees and subset selection in linear regression show that bagging can give substantial gains in accuracy.
IEEE Transactions on Knowledge and Data EngineeringToward the next generation of recommender systems: a survey of the state-of-the-art and possible extensions
10,115 Citations2005Gediminas Adomavičius, Alexander Tuzhilin
This paper presents an overview of the field of recommender systems and describes the current generation of recommendation methods that are usually classified into the following three main categories: content-based, collaborative, and hybrid recommendation approaches.
ACM Transactions on Information SystemsEvaluating collaborative filtering recommender systems
5,762 Citations2004Jonathan L. Herlocker, Joseph A. Konstan +2 more
The key decisions in evaluating collaborative filtering recommender systems are reviewed: the user tasks being evaluated, the types of analysis and datasets being used, the ways in which prediction quality is measured, the evaluation of prediction attributes other than quality, and the user-based evaluation of the system as a whole.
Minds at UW (University of Wisconsin)Active Learning Literature Survey
4,776 Citations2009Burr Settles
This report provides a general introduction to active learning and a survey of the literature, including a discussion of the scenarios in which queries can be formulated, and an overview of the query strategy frameworks proposed in the literature to date.
ACM SIGIR ForumAn Algorithmic Framework for Performing Collaborative Filtering
2,552 Citations2017Jonathan L. Herlocker, Joseph A. Konstan +2 more
Methods and metrics for cold-start recommendations
1,707 Citations2002Andrew I. Schein, Alexandrin Popescul +2 more
Choice Reviews OnlineStatistics for experimenters: design, innovation, and discovery
1,662 Citations2006
Query by committee
1,657 Citations1992H. Sebastian Seung, Manfred Opper +1 more
It is suggested that asymptotically finite information gain may be an important characteristic of good query algorithms, in which a committee of students is trained on the same data set.
Journal of Artificial Intelligence ResearchActive Learning with Statistical Models
1,286 Citations1996David Cohn, Zoubin Ghahramani +1 more
This work shows how the same principles may be used to select data for two alternative, statistically-based learning architectures: mixtures of Gaussians and locally weighted regression.
Being accurate is not enough
1,055 Citations2006Sean M. McNee, John Riedl +1 more
This paper proposes informal arguments that the recommender community should move beyond the conventional accuracy metrics and their associated experimental methodologies, and proposes new user-centric directions for evaluating recommender systems.
An analysis of active learning strategies for sequence labeling tasks
992 Citations2008Burr Settles, Mark Craven
This paper surveys previously used query selection strategies for sequence models, and proposes several novel algorithms to address their shortcomings, and conducts a large-scale empirical comparison.
ComputerGames with a Purpose
959 Citations2006L. von Ahn
"Games with a purpose" have a vast range of applications in areas as diverse as security, computer vision, Internet accessibility, adult content filtering, and Internet search, and any game designed to address these and other problems must ensure that game play results in a correct solution and, at the same time, is enjoyable.
Wiley series in probability and statisticsDesign and Analysis of Experiments
956 Citations2005Klaus Hinkelmann, Oscar Kempthorne
Wiley Encyclopedia of Electrical and Electronics EngineeringComputational Linguistics
834 Citations1999Fred Popowich
Toward Optimal Active Learning through Sampling Estimation of Error Reduction
824 Citations2001Nicholas Roy, Andrew McCallum
Less is More: Active Learning with Support Vector Machines
775 Citations2000Greg Schohn, David Cohn
A simple active learning heuristic is described which greatly enhances the generalization behavior of support vector machines (SVMs) on several practical document classification tasks and frequently does so in less time than the naive approach of training on all available data.
Employing EM and Pool-Based Active Learning for Text Classification
681 Citations1998Andrew McCallum, Kamal Nigam
The Query-by-Committee method of active learning is modified to use the unlabeled pool for explicitly estimating document density when selecting examples for labeling, and the combination of EM and active learning requires only slightly more than half as many labeled training examples to achieve the same accuracy as either the improved active learning or EM alone.
Getting to know you
537 Citations2002Al Mamunur Rashid, István Albert +5 more
Six techniques that collaborative filtering recommender systems can use to learn about new users are studied, showing that the choice of learning technique significantly affects the user experience, in both the user effort and the accuracy of the resulting predictions.
Elsevier eBooksCommittee-Based Sampling For Training Probabilistic Classifiers
497 Citations1995Ido Dagan, Sean P. Engelson
A general method for efficiently training probabilistic classifiers, by selecting for training only the more informative examples in a stream of unlabeled examples, which is particularly attractive because it evaluates the expected information gain from a training example implicitly.
Multiple-Instance Active Learning
464 Citations2007Burr Settles, Mark Craven +1 more
The experiments show that learning from instance labels can significantly improve performance of a basic MI learning algorithm in two multiple-instance domains: content-based image retrieval and text classification.
Collaborative filtering via gaussian probabilistic latent semantic analysis
363 Citations2003Thomas Hofmann
A new model-based algorithm is described, based on a generalization of probabilistic latent semantic analysis to continuous-valued response variables, which assumes that the observed user ratings can be modeled as a mixture of user communities or interest groups.
Learning on the border
362 Citations2007Şeyda Ertekin, Jian Huang +2 more
It is demonstrated that active learning is capable of solving the class imbalance problem by providing the learner more balanced classes and an efficient way of selecting informative instances from a smaller pool of samples for active learning which does not necessitate a search through the entire dataset.
Active learning via transductive experimental design
353 Citations2006Kai Yu, Jinbo Bi +1 more
This paper considers the problem of selecting the most informative experiments x to get measurements y for learning a regression model y, and proposes a novel and simple concept for active learning, transductive experimental design, that has a better scalability in comparison with classic experimental design methods.
Agnostic active learning
282 Citations2006Maria-Florina Balcan, Alina Beygelzimer +1 more
The first active learning algorithm which works in the presence of arbitrary forms of noise is state and analyzed, and it is shown that A2 achieves an exponential improvement over the usual sample complexity of supervised learning.
TechnometricsD-Optimality for Regression Designs: A Review
279 Citations1975Ralph C. St. John, Norman R. Draper
Neural NetworksNeural Network Exploration Using Optimal Experiment Design
263 Citations1996David Cohn
Active Learning with Real Annotation Costs
204 Citations2008Burr Settles, Mark Craven +1 more
A detailed empirical study of active learning with annotation costs in four real-world domains involving human annotators is presented, to better understand the nature of actual labeling costs in domains where labeling costs may vary.
IEEE Intelligent SystemsAcquiring and Revising Preferences in a Critique-Based Mobile Recommender System
174 Citations2007Francesco Ricci⋆, Quang Nhat Nguyen
A critique-based recommendation methodology aids the acquisition and revision of user preferences in a mobile recommender system and is implemented in MobyRek, a mobile-phone recommenders system that helps users search for travel products.
Artificial IntelligenceLearning cost-sensitive active classifiers☆☆This extends the short conference paper [19].
164 Citations2002Russell Greiner, Adam J. Grove +1 more
This paper analyzes the problem of learning optimal active classifiers, using a variant of the probably-approximately-correct (PAC) model, and shows that this task can be achieved efficiently when the active classifier is allowed to perform only (at most) a constant number of tests.
Interactive collaborative filtering
161 Citations2013Xiaoxue Zhao, Weinan Zhang +1 more
This paper forms the interactive CF with the probabilistic matrix factorization (PMF) framework, and leverage several exploitation-exploration algorithms to select items, including the empirical Thompson sampling and upper confidence bound based algorithms.
Real-time top-n recommendation in social streams
155 Citations2012Ernesto Diaz-Aviles, Lucas Drumond +2 more
This paper focuses on analyzing social streams in real-time for personalized topic recommendation and discovery and presents Stream Ranking Matrix Factorization - RMFX, which uses a pairwise approach to matrix factorization in order to optimize the personalized ranking of topics.
Lecture notes in computer scienceOn the Role of Diversity in Conversational Recommender Systems
147 Citations2007Lorraine McGinty, Barry Smyth
This paper describes and fully evaluate a powerful new diversity-enhancing technique that has the ability to significantly improve the performance of conversational recommender systems across the board.
Journal of Machine Learning ResearchActive Learning in Approximately Linear Regression Based on Conditional Expectation of Generalization Error
128 Citations2006Masashi Sugiyama
This paper proposes a new active learning method also using the weighted least-squares learning, which it proves that the proposed active learning criterion is a more accurate predictor of the single-trial generalization error than the existing criterion.
Personalized active learning for collaborative filtering
122 Citations2008Abhay Harpale, Yiming Yang
This paper proposes an extended form of Bayesian active learning and uses the Aspect Model for CF to illustrate and examine the idea, and proposes a method to personalize active learning for the user, and query for only those items which the user can provide rating for.
International Joint Conference on Artificial IntelligenceSelective supervision: guiding supervised learning with decision-theoretic active learning
121 Citations2007Ashish Kapoor, Eric Horvitz +1 more
This work looks at methods that guide the allocation of human effort for labeling data so as to get the greatest boosts in discriminatory power with increasing amounts of work and focuses on the application of value of information to Gaussian Process classifiers.
arXiv (Cornell University)Selective Sampling for Example-based Word Sense Disambiguation
114 Citations1999Atsushi Fujii, Takenobu Tokunaga +2 more
This paper proposes an efficient example sampling method for example-based word sense disambiguation systems, characterized by the reliance on the notion of training utility: the degree to which each example is informative for future example sampling when used for the training of the system.
Uncertainty in Artificial IntelligenceA Bayesian approach toward active learning for collaborative filtering
105 Citations2004Rong Jin, Luo Si
Proceedings of the AAAI Conference on Artificial IntelligenceActive Transfer Learning for Cross-System Recommendation
96 Citations2013Lili Zhao, Sinno Jialin Pan +4 more
This paper proposes a framework to construct entity correspondence with limited budget by using active learning to facilitate knowledge transfer across recommender systems, and first iteratively select entities in the target system based on a proposed criterion to query their correspondences in the source system.
Towards more conversational and collaborative recommender systems
90 Citations2003Giuseppe Carenini, Jocelyin Smith +1 more
This paper proposes a set of techniques to intelligently select what information to elicit from the user in situations in which the user may be particularly motivated to provide such information, and argues that the resulting interaction improves the user experience.
The influence limiter
88 Citations2007Paul Resnick, Rahul Sami
An influence-limiting algorithm that can turn existing recommender systems into manipulation-resistant systems is described and both the influence limits and the information loss incurred due to those limits are described in terms of information-theoretic concepts of loss functions and entropies.
AI MagazineUser‐Involved Preference Elicitation for Product Search and Recommender Systems
87 Citations2008Pearl Pu, Li Chen
Some analyses of common areas of design pitfalls are provided and a set of design guidelines are derived that assist the user in avoiding these problems in three important areas: user preference elicitation, preference revision, and explanation interfaces.
ACM Transactions on Intelligent Systems and TechnologyActive learning strategies for rating elicitation in collaborative filtering
72 Citations2013Mehdi Elahi, Francesco Ricci⋆ +1 more
It is shown that an elicited rating has effects across the system, so a typical user-centric evaluation which ignores any changes of rating prediction of other users also ignores these cumulative effects, which may be more influential on the performance of the system as a whole.
Influence in Ratings-Based Recommender Systems: An Algorithm-Independent Approach
70 Citations2005Al Mamunur Rashid, George Karypis +1 more
This work proposes an algorithm-independent denition of inuence that can be applied to any ratings-based recommender system, and shows experimentally that inUence may be estimated using simple, inexpensive metrics.
Information Systems ResearchDecision-Centric Active Learning of Binary-Outcome Models
65 Citations2007Maytal Saar‐Tsechansky, Foster Provost
A new approach for active data acquisition that specifically targets decision making is introduced, which departs from traditional active learning by placing emphasis on acquisitions that are more likely to affect decision making and describes two different types of decision-centric techniques.
Lecture notes in computer scienceActive Class Selection
59 Citations2007R. Lomasky, Carla E. Brodley +3 more
In an empirical evaluation, it is shown that for a fixed number of training instances, methods based on increasing class stability outperform methods that seek to maximize class accuracy or that use random sampling.
BioinformaticsChoosing where to look next in a mutation sequence space: Active Learning of informative p53 cancer rescue mutants
54 Citations2007Samuel A. Danziger, Jue Zeng +3 more
Five novel and five existing Active Learning techniques, together with three control methods, were tested on 57 previously unknown p53 cancer rescue mutants for their ability to build classifiers that predict protein function and shows that Active Learning is a useful tool for biomedical research.
Journal of Mathematical Analysis and ApplicationsA-optimality for regression designs
54 Citations1982N. N. Chan
Case amazon
45 Citations2007Juha Leino, Kari‐Jouko Räihä
The results underline the need to develop recommender systems as a whole, and the way the recommendations are shown affects which items get picked, and for improving the interface, it is necessary to study the whole in addition to studying the parts in isolation.
Evaluating the dynamic properties of recommendation algorithms
43 Citations2010Robin Burke
A new evaluation method for the dynamic aspects of collaborative algorithms, the "temporal leave-one-out" approach, which can provide insight into both user-specific and system-level evolution of recommendation behavior.
Active learning and search on low-rank matrices
41 Citations2013Danica J. Sutherland, Barnabás Póczos +1 more
This work presents a general approach for active collaborative prediction with the Probabilistic Matrix Factorization model, using variational approximations or Markov chain Monte Carlo sampling to estimate the posterior distribution over models.
Influence-based collaborative active learning
39 Citations2007Neil Rubens, Masashi Sugiyama
This work proposes to select items that are not only uncertain but are also influential, which will reduce the uncertainty over all items and demonstrate the advantages of the proposed approach.
Accelerating Active Learning with Transfer Learning
35 Citations2013David C. Kale, Yan Liu
This paper derives an intuitive bound on generalization error for the classifiers learned by this algorithm that provides insight into the algorithm's behavior and the problem in general and presents a simple and principled transfer active learning framework.
Information Systems ResearchSelling vs. Profiling: Optimizing the Offer Set in Web-Based Personalization
34 Citations2014Monica Johar, Vijay Mookerjee +1 more
One of the most important insights of this study is that frequent visits to the firm's website are extremely important for an e-tailing firm even though the customer may not always buy products during these visits.
International Journal of Pattern Recognition and Artificial IntelligenceSELECTIVE VOTING — GETTING MORE FOR LESS IN SENSOR FUSION
33 Citations2006Lior Rokach, Oded Maimon +1 more
Using "selective voting", the number of sensors is decreased significantly while the performance of the integrated model's classification is increased, which is designed to combine the classification of the models (sensor fusion).
Supporting product selection with query editing recommendations
31 Citations2007Derek Bridge, Francesco Ricci⋆
If the user is given advice, the number of queries she needs to try before finding the product of highest utility is greatly reduced and an advisor that confines its advice to queries that the user model predicts are likely to be tried next will give shorter advice than one whose advice is unconstrained by the user models.
Selectively acquiring ratings for product recommendation
30 Citations2007Zan Huang
This paper proposes proactively acquiring ratings from customers for a newly introduced product to quickly improve the accuracy of the predicted ratings generated by a collaborative filtering recommendation algorithm for the entire customer population.
Neural NetworksA batch ensemble approach to active learning with model selection
28 Citations2008Masashi Sugiyama, Neil Rubens
This paper proposes a new approach called ensemble active learning for solving the problems of active learning and model selection at the same time and demonstrates by numerical experiments that the proposed method compares favorably with alternative approaches such as iteratively performing active learningand model selection in a sequential manner.
Data Mining and Knowledge DiscoveryPessimistic cost-sensitive active learning of decision trees for profit maximizing targeting campaigns
27 Citations2008Lior Rokach, Lihi Naamani +1 more
Pessimistic Active Learning employs a novel pessimistic measure, which relies on confidence intervals and is used to balance the exploration/exploitation trade-off and is compared to those of other methods to indicate PAL’s superiority.
Exploiting the characteristics of matrix factorization for active learning in recommender systems
23 Citations2012Rasoul Karimi, Christoph Freudenthaler +2 more
A method is proposed that improves the most popular selection strategy using the characteristics of matrix factorization and finds similar users to the new user in the latent space and then selects item which is most popular among the similar users.
Machine LearningA Theoretical Analysis of Query Selection for Collaborative Filtering
18 Citations2003Sanjoy Dasgupta, Wee Sun Lee +1 more
It is proved that no polynomial-time algorithm can have a significantly better bound on the number of queries unless all problems in NP have nO(log log n) time algorithms.
IPSJ Online TransactionsOutput Divergence Criterion for Active Learning in Collaborative Settings
13 Citations2009Neil Rubens, Ryota Tomioka +1 more
A new active learning criterion is proposed that is aimed at directly improving the accuracy of the output value estimation by analyzing the effect of the new training points on the estimates of the output values.
The MIT Press eBooksA Conditional Expectation Approach to Model Selection and Active Learning under Covariate Shift
5 Citations2008Masashi Sugiyama, Neil Rubens +1 more
This chapter contains sections titled: Conditional Expectation Analysis of Generalization Error, Linear Regression under Covariate Shift, Model Selection, Active Learning, Active learning with Model selection, Conclusions.
Active Learning for Co-Clustering Based Collaborative Filtering
2 Citations2010Quang Thang Le, Minh Phuong Tu
This paper proposes combining expected value of rating information with likelihood of getting ratings from the users to form the sample selection criterion and shows that the proposed method outperforms three popular active learning strategies for collaborative filtering.