Preference-based learning to rank

被引:18
作者
Ailon, Nir [1 ]
Mohri, Mehryar [1 ,2 ]
机构
[1] Technion Israel Inst Technol, Fac Comp Sci, IL-32000 Haifa, Israel
[2] NYU, Courant Inst Math Sci, New York, NY 10012 USA
关键词
Learning to rank; Machine learning reductions; ROC; ROC CURVE; AREA;
D O I
10.1007/s10994-010-5176-9
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
This paper presents an efficient preference-based ranking algorithm running in two stages. In the first stage, the algorithm learns a preference function defined over pairs, as in a standard binary classification problem. In the second stage, it makes use of that preference function to produce an accurate ranking, thereby reducing the learning problem of ranking to binary classification. This reduction is based on the familiar QuickSort and guarantees an expected pairwise misranking loss of at most twice that of the binary classifier derived in the first stage. Furthermore, in the important special case of bipartite ranking, the factor of two in loss is reduced to one. This improved bound also applies to the regret achieved by our ranking and that of the binary classifier obtained. Our algorithm is randomized, but we prove a lower bound for any deterministic reduction of ranking to binary classification showing that randomization is necessary to achieve our guarantees. This, and a recent result by Balcan et al., who show a regret bound of two for a deterministic algorithm in the bipartite case, suggest a trade-off between achieving low regret and determinism in this context. Our reduction also admits an improved running time guarantee with respect to that deterministic algorithm. In particular, the number of calls to the preference function in the reduction is improved from Omega(n (2)) to O(nlog n). In addition, when the top k ranked elements only are required (ka parts per thousand(a)n), as in many applications in information extraction or search engine design, the time complexity of our algorithm can be further reduced to O(klog k+n). Our algorithm is thus practical for realistic applications where the number of points to rank exceeds several thousand.
引用
收藏
页码:189 / 211
页数:23
相关论文
共 50 条
  • [1] Preference-based learning to rank
    Nir Ailon
    Mehryar Mohri
    Machine Learning, 2010, 80 : 189 - 211
  • [2] Preference-based Learning of Ideal Solutions in TOPSIS-like Decision Models
    Agarwal, Manish
    Tehrani, Ali Fallah
    Huellermeier, Eyke
    JOURNAL OF MULTI-CRITERIA DECISION ANALYSIS, 2015, 22 (3-4) : 175 - 183
  • [3] Preference Learning to Rank with Sparse Bayesian
    Chang, Xiao
    Zheng, Qinghua
    2009 IEEE/WIC/ACM INTERNATIONAL JOINT CONFERENCES ON WEB INTELLIGENCE (WI) AND INTELLIGENT AGENT TECHNOLOGIES (IAT), VOL 3, 2009, : 143 - 146
  • [4] A pairwise learning to rank algorithm based on bounded loss and preference weight
    Tang, Xianlun
    Xiong, Deyi
    Li, Jiaxin
    Wan, Yali
    2017 CHINESE AUTOMATION CONGRESS (CAC), 2017, : 7025 - 7029
  • [5] SortNet: Learning to Rank by a Neural Preference Function
    Rigutini, Leonardo
    Papini, Tiziano
    Maggini, Marco
    Scarselli, Franco
    IEEE TRANSACTIONS ON NEURAL NETWORKS, 2011, 22 (09): : 1368 - 1380
  • [6] A preference-based measure for test performance with an application to prenatal diagnostics
    Felder, Stefan
    Robra, Berrit-Peter
    STATISTICS IN MEDICINE, 2006, 25 (21) : 3696 - 3706
  • [7] RankCNN: When learning to rank encounters the pseudo preference feedback
    Dong, Yuan
    Huang, Chong
    Liu, Wei
    COMPUTER STANDARDS & INTERFACES, 2014, 36 (03) : 554 - 562
  • [8] Multiobjective Learning to Rank Based on the (1
    Ismail, Walaa N.
    Ibrahim, Osman Ali Sadek
    Alsalamah, Hessah A.
    Mohamed, Ebtesam
    ELECTRONICS, 2023, 12 (17)
  • [9] Learning To Rank Based on Modified Genetic Algorithm
    Semenikhin, S. V.
    Denisova, L. A.
    2016 DYNAMICS OF SYSTEMS, MECHANISMS AND MACHINES (DYNAMICS), 2016,
  • [10] Learning to Rank based Gene Summary Extraction
    Shang, Yue
    Hao, Huihui
    Lin, Hongfei
    Wu, Jiajin
    2013 IEEE INTERNATIONAL CONFERENCE ON BIOINFORMATICS AND BIOMEDICINE (BIBM), 2013,