Filters








38,421 Hits in 3.9 sec

Ranking function adaptation with boosting trees

Keke Chen, Jing Bai, Zhaohui Zheng
2011 ACM Transactions on Information Systems  
Tree adaptation assumes that ranking functions are trained with the Stochastic Gradient Boosting Trees method − a gradient boosting method on regression trees.  ...  In this paper, we propose a new approach called tree based ranking function adaptation ("Trada") to effectively utilize these data sources for training cross-domain ranking functions.  ...  Although it can be applied to any regression-tree based ranking model, we will use rank-ing functions trained with the gradient boosting trees (GBT) method [Friedman 2001] in this paper.  ... 
doi:10.1145/2037661.2037663 fatcat:lbhweoqkzve7phaxrpfse3utbe

Model Adaptation via Model Interpolation and Boosting for Web Search Ranking [article]

Jianfeng Gao, Qiang Wu, Chris Burges, Krysta Svore, Yi Su, Nazan Khan, Shalin Shah, Hongyan Zhou
2019 arXiv   pre-print
This paper explores two classes of model adaptation methods for Web search ranking: Model Interpolation and error-driven learning approaches based on a boosting algorithm.  ...  The tree-based boosting algorithm achieves the best performance on most of the closed test sets where the test data and the training data are similar, but its performance drops significantly on the open  ...  We thank Steven Yao's group at Microsoft Bing Search for their help with the experiments.  ... 
arXiv:1907.09471v1 fatcat:t3yo5aomxnhh3nooyjijtblvgq

Adapting boosting for information retrieval measures

Qiang Wu, Christopher J. C. Burges, Krysta M. Svore, Jianfeng Gao
2009 Information retrieval (Boston)  
In addition, we show that starting with a previously trained model, and boosting using its residuals, furnishes an effective technique for model adaptation, and we give significantly improved results for  ...  We present a new ranking algorithm that combines the strengths of two previous methods: boosted tree classification, and LambdaRank, which has been shown to be empirically optimal for a widely used information  ...  MART is a boosted tree algorithm that performs gradient descent in function space [18] .  ... 
doi:10.1007/s10791-009-9112-1 fatcat:5i6mlb2gqveqljo55y3axblq2e

Multi-task learning for boosting with application to web search ranking

Olivier Chapelle, Pannagadatta Shivaswamy, Srinivas Vadrevu, Kilian Weinberger, Ya Zhang, Belle Tseng
2010 Proceedings of the 16th ACM SIGKDD international conference on Knowledge discovery and data mining - KDD '10  
In this paper we propose a novel algorithm for multi-task learning with boosted decision trees.  ...  We evaluate our learning method on web-search ranking data sets from several countries.  ...  As already noted in the introduction, boosted decision trees are very well suited for our web search ranking problem and we now present our algorithm, multi-boost for multi-task learning with boosting.  ... 
doi:10.1145/1835804.1835953 dblp:conf/kdd/ChapelleSVWZT10 fatcat:36ewsgphr5dohmy7k4xhk4mvuu

Boosted multi-task learning

Olivier Chapelle, Pannagadatta Shivaswamy, Srinivas Vadrevu, Kilian Weinberger, Ya Zhang, Belle Tseng
2010 Machine Learning  
In this paper we propose a novel algorithm for multi-task learning with boosted decision trees.  ...  Our algorithm is derived using the relationship between 1 -regularization and boosting. We evaluate our learning method on web-search ranking data sets from several countries.  ...  As already noted in the introduction, boosted decision trees are very well suited for our web search ranking problem and we now present our algorithm, multi-boost for multi-task learning with boosting.  ... 
doi:10.1007/s10994-010-5231-6 fatcat:7r6t3zwlk5d3hirxfjkuur4qq4

InfiniteBoost: building infinite ensembles with gradient descent [article]

Alex Rogozhnikov, Tatiana Likhomanenko
2018 arXiv   pre-print
Two notable ensemble methods widely used in practice are gradient boosting and random forests.  ...  of trees without the over-fitting effect.  ...  For ranking task the fixed capacity value is used because the loss function is not convex in this case and the adaptation on the holdout significantly underestimates capacity.  ... 
arXiv:1706.01109v2 fatcat:z3kxlhjqcnhixayj52h7lm4gya

Trada

Keke Chen, Rongqing Lu, C. K. Wong, Gordon Sun, Larry Heck, Belle Tseng
2008 Proceeding of the 17th ACM conference on Information and knowledge mining - CIKM '08  
Tree adaptation assumes that ranking functions are trained with regression-tree based modeling methods, such as Gradient Boosting Trees.  ...  In this paper, we propose a new approach called tree based ranking function adaptation ("tree adaptation") to address this problem.  ...  Although it can be applied to any regression-tree based ranking models, we will use ranking functions trained with the gradient boosting trees (GBT) method [10] in this paper.  ... 
doi:10.1145/1458082.1458233 dblp:conf/cikm/ChenLWSHT08 fatcat:5o3gpy6ambc2vot2jfubpnvoua

Active learning of tree tensor networks using optimal least-squares [article]

Cécile Haberstich, Anthony Nouy, Guillaume Perrin
2021 arXiv   pre-print
Practical strategies are proposed for adapting the feature spaces and ranks to achieve a prescribed error.  ...  In this paper, we propose new learning algorithms for approximating high-dimensional functions using tree tensor networks in a least-squares setting.  ...  D Estimation of the α-ranks of a function u to perform tree adaptation. We present here the algorithm that estimates α-ranks for tree adaptation. The strategy is described in Section 4.1.  ... 
arXiv:2104.13436v1 fatcat:3wzbu57hujcbbelf2ipdhyiini

Review of statistical methods for survival analysis using genomic data

Seungyeoun Lee, Heeju Lim
2019 Genomics & Informatics  
adapted to survival analysis.  ...  We review traditional survival methods and regularization methods, with various penalty functions, for the analysis of high-dimensional genomics, and describe machine learning techniques that have been  ...  Both mboost and Cox-Boost are based on gradient boosting, but differ in the sense that mboost is an adaptation of model-based boosting, whereas Cox-Boost adapts likelihood-based boosting.  ... 
doi:10.5808/gi.2019.17.4.e41 pmid:31896241 pmcid:PMC6944043 fatcat:dw7rubh7v5a3hcgptsyqnydk6a

McRank: Learning to Rank Using Multiple Classification and Gradient Boosting

Ping Li, Christopher J. C. Burges, Qiang Wu
2007 Neural Information Processing Systems  
We propose using the Expected Relevance to convert class probabilities into ranking scores. The class probabilities are learned using a gradient boosting tree algorithm.  ...  function, although the reported results in [5] are for pairwise.  ...  Regression-based Ranking Using Boosting Tree Algorithm With slight modifications, the boosting tree algorithm can be used for regressions.  ... 
dblp:conf/nips/LiBW07 fatcat:hj7zoqzcofcidjiiamtfpvxojq

Gradient Boosting Machine: A Survey [article]

Zhiyuan He, Danchen Lin, Thomas Lau, Mike Wu
2019 arXiv   pre-print
optimization, 3. loss function estimations, and 4. model constructions. 5. application of boosting in ranking.  ...  In this survey, we discuss several different types of gradient boosting algorithms and illustrate their mathematical frameworks in detail: 1. introduction of gradient boosting leads to 2. objective function  ...  Meanwhile, boosting performance depends on bias reduction when the weighted sampling is replaced with weighted tree fitting (Friedman et al., 2000) .  ... 
arXiv:1908.06951v1 fatcat:fgofwpdrn5hptfdqv2bzgbwcou

Gradient Boosting Neural Networks: GrowNet [article]

Sarkhan Badirli, Xuanqing Liu, Zhengming Xing, Avradeep Bhowmik, Khoa Doan, Sathiya S. Keerthi
2020 arXiv   pre-print
General loss functions are considered under this unified framework with specific examples presented for classification, regression, and learning to rank.  ...  A fully corrective step is incorporated to remedy the pitfall of greedy function approximation of classic gradient boosting decision tree.  ...  The adaptive boosting can be seen as a specific version of the gradient boosting algorithm where a simple exponential loss function is used [10] .  ... 
arXiv:2002.07971v2 fatcat:ck4smv5vrne7bf2f4crl7lxgei

On domain similarity and effectiveness of adapting-to-rank

Keke Chen, Jing Bai, Srihari Reddy, Belle Tseng
2009 Proceeding of the 18th ACM conference on Information and knowledge management - CIKM '09  
Adapting to rank address the the problem of insufficient domainspecific labeled training data in learning to rank. However, the initial study shows that adaptation is not always effective.  ...  In this paper, we investigate the relationship between the domain similarity and the effectiveness of domain adaptation with the help of two domain similarity measure: relevance correlation and sample  ...  the source domain function to the target domain with the Trada tree adaptation algorithm [1] that adjusts the gradient boosting tree structure with the target domain data.  ... 
doi:10.1145/1645953.1646182 dblp:conf/cikm/ChenBRT09 fatcat:kbcvqwag2ffkjhx6q46bmyshge

Plackett-Luce model for learning-to-rank task [article]

Tian Xia, Shaodan Zhai, Shaojun Wang
2019 arXiv   pre-print
List-wise based learning to rank methods are generally supposed to have better performance than point- and pair-wise based.  ...  Gradient Boosting and Regression Tree We review gradient boosting [14] as a general framework for function approximation using regression trees as the weak learners, which has been the most successful  ...  A ranking function f scores each query-document pair, and returns sorted documents associated with the same query.  ... 
arXiv:1909.06722v1 fatcat:5rrqzb5vxfbzplvy2jrwgu44wm

Real-Time Face Identification via CNN and Boosted Hashing Forest

Yury Vizilter, Vladimir Gorbatsevich, Andrey Vorotnikov, Nikita Kostromov
2016 2016 IEEE Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)  
This BHF generalizes the Boosted SSC approach for hashing learning with joint optimization of face verification and identification.  ...  The family of real-time face representations is obtained via Convolutional Network with Hashing Forest (CNHF).  ...  Our CNHF with 2000 output 7-bit coding trees (CNHF-2000×7) achieves 98.59% verification accuracy and 93% rank-1 on LFW (add 3% to rank-1 of basic CNN).  ... 
doi:10.1109/cvprw.2016.25 dblp:conf/cvpr/VizilterGVK16 fatcat:5wjpjs4rpndn7c3misu5eils7u
« Previous Showing results 1 — 15 out of 38,421 results