Relevance feedback and category search in image databases

Meilhac, Christophe; Nastar, Chahab

doi:10.1109/mmcs.1999.779254

Cited by 75 publications

(40 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…In [23], the authors propose a Bayesian model which supports image classes that assign a high membership probability to PE images and penalizes classes that assign a high membership probability to NE images. In [24], Meilhac et al consider that the image collection is made up of relevant images, among which the user chooses the PE, and nonrelevant images, among which the user chooses the NE. They use a Bayesian model in which they try to estimate the distribution of relevant images and simultaneously minimize the probability of retrieving nonrelevant images.…”

Section: A Overview Of the State Of The Artmentioning

confidence: 99%

Relevance feedback for CBIR: a new approach based on probabilistic feature weighting with positive and negative examples

Kherfi

Ziou

2006

IEEE Trans. on Image Process.

View full text Add to dashboard Cite

Abstract-In content-based image retrieval, understanding the user's needs is a challenging task that requires integrating him in the process of retrieval. Relevance feedback (RF) has proven to be an effective tool for taking the user's judgement into account. In this paper, we present a new RF framework based on a feature selection algorithm that nicely combines the advantages of a probabilistic formulation with those of using both the positive example (PE) and the negative example (NE). Through interaction with the user, our algorithm learns the importance he assigns to image features, and then applies the results obtained to define similarity measures that correspond better to his judgement. The use of the NE allows images undesired by the user to be discarded, thereby improving retrieval accuracy. As for the probabilistic formulation of the problem, it presents a multitude of advantages and opens the door to more modeling possibilities that achieve a good feature selection. It makes it possible to cluster the query data into classes, choose the probability law that best models each class, model missing data, and support queries with multiple PE and/or NE classes. The basic principle of our algorithm is to assign more importance to features with a high likelihood and those which distinguish well between PE classes and NE classes. The proposed algorithm was validated separately and in image retrieval context, and the experiments show that it performs a good feature selection and contributes to improving retrieval effectiveness.Index Terms-Content-based image retrieval (CBIR), feature selection (FS), relevance feedback (RF).

show abstract

Section: A Overview Of the State Of The Artmentioning

confidence: 99%

Relevance feedback for CBIR: a new approach based on probabilistic feature weighting with positive and negative examples

Kherfi

Ziou

2006

IEEE Trans. on Image Process.

View full text Add to dashboard Cite

show abstract

“…Problems with unpredictable performance similar to the vector space model still persist. Nastar et al [15] use two non-parametric density estimates to model the probability distribution for the relevant and non-relevant classes. The relevance score of a given database object is defined as,…”

Section: Related Workmentioning

confidence: 99%

Adaptable Similarity Search using Non-Relevant Information

Ashwin

Gupta

Ghosal

2002

VLDB '02: Proceedings of the 28th International Conference on Very Large Databases

View full text Add to dashboard Cite

Many modern database applications require content-based similarity search capability in numeric attribute space. Further, users' notion of similarity varies between search sessions. Therefore online techniques for adaptively refining the similarity metric based on relevance feedback from the user are necessary. Existing methods use retrieved items marked relevant by the user to refine the similarity metric, without taking into account the information about non-relevant (or unsatisfactory) items. Consequently items in database close to non-relevant ones continue to be retrieved in further iterations. In this paper a robust technique is proposed to incorporate non-relevant information to efficiently discover the feasible search region. A decision surface is determined to split the attribute space into relevant and non-relevant regions. The decision surface is composed of hyperplanes, each of which is normal to the minimum distance vector from a nonrelevant point to the convex hull of the relevant points. A similarity metric, estimated using the relevant objects is used to rank and retrieve database objects in the relevant region. Experiments on simulated and benchmark datasets demonstrate robustness and superior performance of the proposed technique over existing adaptive similarity search techniques.

show abstract

“…Some algorithm assumes the user will give a binary feedback for positive and negative examples [29]; some only takes positive examples [12]; some takes positive and negative examples with "degree of (ir)relevance" for each [21]; some assumes the feedback is only a comparative judgment, i.e., the positive examples are not necessarily "relevant" to the target, but "more like the target than the negative ones" [6]. The latter can be related to "query refinement" techniques in others [14].…”

Section: Variantsmentioning

confidence: 99%

Comparing discriminating transformations and SVM for learning during multimedia retrieval

Zhou

Huang

2001

Proceedings of the Ninth ACM International Conference on Multimedia

View full text Add to dashboard Cite

On-line learning or "relevance feedback" techniques for multimedia information retrieval have been explored from many different points of view: from early heuristic-based feature weighting schemes to recently proposed optimal learning algorithms, probabilistic/Bayesian learning algorithms, boosting techniques, discriminant-EM algorithm, support vector machine, and other kernel-based learning machines. Based on a careful examination of the problem and a detailed analysis of the existing solutions, we propose several discriminating transforms as the learning machine during the user interaction. We argue that relevance feedback problem is best represented as a biased classification problem, or a (1+x)-class classification problem. Biased Discriminant Transform (BDT) is shown to outperform all the others. A kernel form is proposed to capture non-linearity in the class distributions.

show abstract

Relevance feedback and category search in image databases

Cited by 75 publications

References 11 publications

Relevance feedback for CBIR: a new approach based on probabilistic feature weighting with positive and negative examples

Relevance feedback for CBIR: a new approach based on probabilistic feature weighting with positive and negative examples

Adaptable Similarity Search using Non-Relevant Information

Comparing discriminating transformations and SVM for learning during multimedia retrieval

Contact Info

Product

Resources

About