Information-theoretic metric learning

Davis, Jason V.; Kulis, Brian; Jain, Prateek; Sra, Suvrit; Dhillon, Inderjit S.

doi:10.1145/1273496.1273523

Cited by 1,637 publications

(1,512 citation statements)

References 9 publications

Supporting

Mentioning

1,504

Contrasting

Unclassified

Order By: Relevance

“…We use the default settings for c and l in the authors' code [12]. The setting of K determines "how local" the learner is; its optimal setting depends on the training data and query.…”

Section: Methodsmentioning

confidence: 99%

See 1 more Smart Citation

Fine-Grained Visual Comparisons with Local Learning

Grauman

2014

2014 IEEE Conference on Computer Vision and Pattern Recognition

419

302

View full text Add to dashboard Cite

Given two images, we want to predict which exhibits a particular visual attribute more than the other-even when the two images are quite similar. Existing relative attribute methods rely on global ranking functions; yet rarely will the visual cues relevant to a comparison be constant for all data, nor will humans' perception of the attribute necessarily permit a global ordering. To address these issues, we propose a local learning approach for fine-grained visual comparisons. Given a novel pair of images, we learn a local ranking model on the fly, using only analogous training comparisons. We show how to identify these analogous pairs using learned metrics. With results on three challenging datasets-including a large newly curated dataset for fine-grained comparisons-our method outperforms stateof-the-art methods for relative attribute prediction.

show abstract

“…We use the default settings for c and l in the authors' code [12]. The setting of K determines "how local" the learner is; its optimal setting depends on the training data and query.…”

Section: Methodsmentioning

confidence: 99%

“…We employ the information-theoretic metric learning (ITML) algorithm [12], due to its efficiency and kernelizability. Figure 4: Example fine-grained neighbor pairs for three test pairs (top row) from the datasets tested in this paper.…”

Section: Selecting Fine-grained Neighboring Pairsmentioning

confidence: 99%

Fine-Grained Visual Comparisons with Local Learning

Grauman

2014

2014 IEEE Conference on Computer Vision and Pattern Recognition

419

302

View full text Add to dashboard Cite

show abstract

“…with x and x ′ image indexes and A 0 a positive semi-definite matrix, that can be learnt using optimization [43,11,24] or boosting [17]. The method in [11] is interesting for our context, since the algorithm seems to be able to update matrix A for each new label.…”

Section: Kernel Learningmentioning

confidence: 99%

“…The method in [11] is interesting for our context, since the algorithm seems to be able to update matrix A for each new label. All values of matrix A are subject to change, and thus all distance values between labelled and labelled/unlabelled images must be recomputed.…”

Section: Kernel Learningmentioning

confidence: 99%

Incremental kernel learning for active image retrieval without global dictionaries

Gosselin

Precioso

Philipp‐Foliguet

2011

Pattern Recognition

View full text Add to dashboard Cite

In content-based image retrieval context, a classic strategy consists in computing off-line a dictionary of visual features. This visual dictionary is then used to provide a new representation of the data which should ease any task of classification or retrieval. This strategy, based on past research works in text retrieval, is suitable for the context of batch learning, when a large training set can be built either by using a strong prior knowledge of data semantics (like for textual data) or with an expensive off-line pre-computation. Such an approach has major drawbacks in the context of interactive retrieval, where the user iteratively builds the training data set in a semi-supervised approach by providing positive and negative annotations to the system in the relevance feedback loop. The training set is thus built for each retrieval session without any prior knowledge about the concepts of interest for this session. We propose a completely different approach to build the dictionary on-line from features extracted in relevant images. We design the corresponding kernel function, which is learnt during the retrieval session. For each new label, the kernel function is updated with a complexity linear with respect to the size of the database. We propose an efficient active learning strategy for the weakly supervised retrieval method developed in this paper. Moreover this framework allows the combination of features of different types. Experiments are carried out on standard databases, and show that a small dictionary can be dynamically extracted from the features with better performances than a global one.

show abstract

“…Formally, the Mahanalobis distance between two data x, y ∈ R d is MD(x, y) = (x − y) T M(x − y), the goal is to learn a proper symmetric positive definite matrix M ∈ R d×d . Information theoretic metric learning (ITML) [6] is one of the state-of-the-art methods for Mahalanobis metric learning which uses an information theoretic approach to optimize M under the constraints that the similarity between each pair labeled "same" is below a specified threshold and the one between each pair labeled "different" is above another specified threshold. Chechik et al [42] learnt a parametric similarity function which gives supervision on the relative similarity between two pairs of images through a bilinear form.…”

Section: Introductionmentioning

confidence: 99%

Face identification using reference-based features with message passing model

Shen

Wang

et al. 2013

Neurocomputing

View full text Add to dashboard Cite

In this paper, we propose a system for face identification. Given two query face images, our task is to tell whether or not they are of the same person. The main contribution of this paper comes from two aspects: (1) We adopt the one-shot similarity kernel [35] for learning the similarity of two face images. The learned similarity measures are then used to map a face image to reference images. (2) We propose a graph-based method for selecting an optimal set of reference images. Instead of directly working on the image features, we use the learned similarity to the reference images as the new features and compute the corresponding matching score of the two query images. Our approach is effective and easy to implement. We show encouraging and favorable results on the "Labeled Faces in the Wild" -a challenging data set of faces.

show abstract

Information-theoretic metric learning

Cited by 1,637 publications

References 9 publications

Fine-Grained Visual Comparisons with Local Learning

Fine-Grained Visual Comparisons with Local Learning

Incremental kernel learning for active image retrieval without global dictionaries

Face identification using reference-based features with message passing model

Contact Info

Product

Resources

About