Efficient algorithms for ranking with SVMs

By Olivier Chapelle, S. Sathiya Keerthi. 2010

287 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Signal Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Signal Processing

Citations in the corpus

Regularized Behavior Value Estimation (2021, methodology): “We adopt the squared-hinge loss which is commonly used for RankSVM (Chapelle and Keerthi, 2010).”
Don't just listen, use your imagination: Leveraging visual common sense for non-visual tasks (2015, methodology): “We learn w using a ranking SVM [7]: (1) Visual paraphrasing.”
Web-Search Ranking with Initialized Gradient Boosted Regression Trees (2010, background): “The past years have seen many different approaches to web search ranking, including adaptations of support vector machines (Joachims, 2002; Chapelle and Keerthi, 2010), neural networks (Burges et al.”
Interestingness Prediction by Robust Learning to Rank (2014, methodology): “After that the prediction models differ – [16] converts pairwise comparisons into an absolute interestingness values and use a regression model, whilst [21] employs rankSVM [3] to learn a ranking function, with the estimated ranking score of an unseen video used as the interestingness prediction.”
Person Re-identification by Descriptive and Discriminative Classification (2011, methodology): “These approaches can mainly be subdivided in two groups: (a) methods which employ a representation of descriptive statistics of the human appearance (using hand crafted features) [4,6,17,22] and (b) approaches that are based on discriminative learning [3,8,13,16,18].”
Automated Facial Trait Judgment and Election Outcome Prediction: Social Dimensions of Face (2015, methodology): “We use the implementation of [8] to solve this optimization problem.”
Relaxed Pairwise Learned Metric for Person Re-identification (2012, unknown): “, [3, 5– 7, 14, 8, 4]), and (c) metric learning methods (e.”
Person Re-Identification by Support Vector Ranking (2010, methodology): “For each˜Z i , we compute a weak ranker w i by using a primal-based RankSVM of Chapelle and Keerthi [1], which is tractable given a moderate size dataset.”
Person Re-identification by Video Ranking (2014, methodology): “9 21.5 30.9 45.2 5.1 14.9 20.7 31.3 MS-SDALF [11] 5.2 20.7 32.0 47.9 6.3 18.8 27.1 37.3 Salience [49] with RankSVM [6] as the distance metric.”
Learning to Rank for Information Retrieval and Natural Language Processing (2011, unknown): “Chapelle & Keerthi [16] have developed an efficient algorithm for training Ranking SVM.”
Online Learning: A Comprehensive Survey (2018, methodology): “The Streaming Distributed Stochastic Gradient Descent algorithm (Ali et al., 2011) provides an operable approach to addresses this problem.”
Visual Persuasion: Inferring Communicative Intents of Images (2014, methodology): “We use the implementation of [3] to solve this optimization problem.”
Learning (predictive) risk scores in the presence of censoring due to interventions (2015, methodology): “(3), we obtain the following unconstrained convex optimization formulation: Instead of solving the dual formulation as in Joachims (2002), following the reasons of efﬁciency and accuracy discussed by Chapelle and Keerthi (2010), we solve the primal form of this optimization program as follows.”
Learning a category independent object detection cascade (2011, background): “1We recover the classical ranking SVM [14, 6] in the case that all losses are in {0, 1} and there is exactly one training sample.”
Deep Learning the City: Quantifying Urban Perception at a Global Scale (2016, methodology): “The RSS-CNN learns an additional set of weights—in comparison to SS-CNN—for minimizing a ranking loss, The ranking loss ( L r ) is designed to penalize the network to satisfy the constraints of our ranking problem—which is identical to the the loss function of the RankSVM [50,53] formulation.”
Activity Auto-Completion: Predicting Human Activities from Partial Videos (2015, methodology): “We solve the ranking function with the publicly available PRSVM software [9] which integrates primal Newton method to speed up RankSVM training and shows excellent performance.”
Yahoo! Learning to Rank Challenge Overview (2010, methodology): “There are several efficient implementation of this algorithm that are publicly available (Joachims, 2006; Chapelle and Keerthi, 2010).”
Substructure counting graph kernels for machine learning from RDF data (2015, methodology): “(18) The advantage of the feature vector representation is that it leads to efficient computations of the kernels [33] and for larger datasets the feature vectors can be used directly as input to linear learning methods, such as linear Support Vector Machines [34].”