Metric Learning to Rank

By Brian McFee, Gert R. G. Lanckriet. 2010

367 citations recorded by the source corpus.

How researchers see this paper

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

1. Liang Zheng
2. Xiaogang Wang
3. Yi Yang
4. Wanli Ouyang
5. Chunhua Shen
6. Dacheng Tao
7. Luc Van Gool
8. Yoshua Bengio

Citations in the corpus

• Weakly Supervised Disentanglement with Guarantees (2019, background): “Although supervision via ranking features prominently in the metric learning literature (McFee & Lanckriet, 2010; Wang et al., 2014), our focus in this paper will be on rank pairing in the context of disentanglement guarantees.”
• A Survey on Metric Learning for Feature Vectors and Structured Data (2013, background): “MLR (McFee & Lankriet) The idea of MLR (McFee and Lanckriet, 2010), for Metric Learning to Rank, is to learn a metric for a ranking task, where given a query instance, one aims at producing a ranked list of examples where relevant ones are ranked higher than irrelevant ones.28 Let P the set of all p”
• From region similarity to category discovery (2011, methodology): “Note that k-nearest neighbor accuAlgorithm 1 Metric Learning to Rank [14] Input: data X = {x1, x2, .”
• Person Re-identification by Salience Matching (2013, background): “The partial order feature [9, 17] is appropriate for our goal and can well encode the difference between relevant pairs and irrelevant pairs with only partial orders.”
• Decomposing fit semantics for product size recommendation in metric spaces (2018, background): “Another study uses users’ meta-data to build representations and employs metric learning to retrieve suitable partners for online dating recommendation [7].”
• Embedding Deep Metric for Person Re-identification: A Study Against Large Variations (2016, methodology): “The traditional methods include LOMO-XQDA [19], KISSME [13], LDM [10], RANK [24], eSDC [40], SDALF [7], LMNN [32], ITML [5], Euclid [40].”
• Joint Learning of Single-Image and Cross-Image Representations for Person Re-identification (2016, methodology): “…performances of the proposed method and some other state-of-the-art methods, including Euclidean distance, ITML [5], LMNN [35], metric learning to rank (RANK) [27], LDML [13], symmetry-driven accumulation of local features (SDALF) [8], eSDC [41], KISSME [16], FPNN [19], the work by Ahmed et al .”
• An Overview of Cross-Media Retrieval: Concepts, Methodologies, Benchmarks, and Challenges (2017, background): “information to achieve better metric results, which are widely used in single-media retrieval [52], [53].”
• Learning Deep Feature Representations with Domain Guided Dropout for Person Re-identification (2016, methodology): “Apart from the feature learning methods, a large number of metric learning algorithms [7, 19, 29, 32, 40, 41] have also been proposed to solve the Re-ID problem from a complementary perspective.”
• Deep content-based music recommendation (2013, methodology): “We also used it as input for the metric learning to rank (MLR) algorithm [21], to learn a similarity metric for content-based recommendation.”
• An improved deep learning architecture for person re-identification (2015, methodology): “We compare our method against KISSME [12], eSDC [30], SDALF [5], ITML [4], logistic distance metric learning (LDM) [9], largest margin nearest neighbor (LMNN) [24], metric learning to rank (RANK) [21], and directly using Euclidean distance to compare features.”
• Unbiased Metric Learning: On the Utilization of Multiple Datasets and Web Images for Softening Bias (2013, methodology): “In this experiment, we compared our approach, UML, to the following baseline methods: Structural Metric Learning (SML) [15], Large Margin Nearest Neighbor (LMNN) [24], Maximally Collapsing Metric Learning (MCML) [9] and Large Margin Multi-Task Metric Learning (mtLMNN) [17].”
• Similarity Comparisons for Interactive Fine-Grained Categorization (2014, methodology): “We instead learn a metric of perceptual similarity for categorization from relative comparisons [33, 1, 24, 34], specifically employing stochastic triplet embedding [36] in this work.”
• DeepReID: Deep Filter Pairing Neural Network for Person Re-identification (2014, methodology): “RANK is optimized for ranking problems, while person re-identification is a ranking problem.”
• Fixed-dimensional acoustic embeddings of variable-length segments in low-resource settings (2013, methodology): “MLR is a largemargin approach that aims to separate vectors that are similar to a given query vector from those that are dissimilar by a margin given by a ranking loss, which in our case is mean average precision Given the learned matrix M , we find a matrix U whose ith row is √ |λi|vi, where vi is ”
• Learning to rank in person re-identification with metric ensembles (2015, methodology): “applies the Metric Learning to Rank (MLR) method of [25] to person re-id [43].”
• Metric Learning: A Survey (2013, methodology): “As an extension beyond the simple nearest-neighbor applications of most metric learning techniques, McFee and Lanckriet [52] present a method for utilizing Mahalanobis metric learning for the problem of ranking.”
• Robustness and generalization for metric learning (2012, methodology): “ived in [19]. However, it is known that sparse algorithms are not stable [21], and thus stability-based analysis fails to assess the generalization ability of recent sparse metric learning approaches [22, 23, 8, 9, 29]. The key advantage of robustness over stability is that we can obtain bounds simi”
• PersonNet: Person Re-identification with Deep Convolutional Neural Networks (2016, background): “Last but not least, learning to rank can be employed in person re-id, and approaches include ensembled RankSVM [37], Metric Learning to Rank (MLR) [38] and its application to person re-id [36] and structured metric ensembles [39].”