Confidence interval approach to feature re-weighting

Abstract

Relevance feedback is commonly incorporated into content-based image retrieval systems with the objective of improving retrieval accuracy via user feedback. One effective method for improving retrieval performance is to perform feature re-weighting based on the obtained feedback. Previous approaches to feature re-weighting via relevance feedback assume the feature data for images can be represented in fixed-length vectors. However, many approaches are invalidated with the recent development of features that cannot be represented in fixed-length vectors. In addition, previous approaches use only the information from the set of images returned in the latest query result for feature reweighting. In this paper, we propose a feature re-weighting approach that places no restriction on the representation of feature data and utilizes the aggregate set of images returned over the iterations of retrieval to obtain feature re-weighting information. The approach analyzes the feature distances calculated between the query image and the resulting set of images to approximate the feature distances for the entire set of images in the database. Two-sided confidence intervals are used with the distances to obtain the information for feature re-weighting. There is no restriction on how the distances are calculated for each feature. This provides freedom for how the feature representations are structured. The experimental results show the effectiveness of the proposed approach and in comparisons with other work, it is shown that our approach outperforms previous work.

Multilevel and multiple approaches for Feature Reweighting to reduce semantic gap using relevance feedback

“…Precision (19) and recall (20) are used to evaluate the performance of the propose approach. Precision is the number of the retrieved relevant images over the total number of retrieved images, and recall is the number of the retrieved relevant images over the total number of relevant images in the database.…”

“…Since the performance also depends on the scope, we conducted experiments on scopes of 20, 40, SO, 120 and 200, respectively. For a comparison, we provide the performance of the MARS [7] [10] [II] approach, and CIA [20] under the same experimental conditions. The following Table.I and Table.II show comparison results of our proposed system with MARS and CIA in terms of precision and recall, it outperforms other two approaches.…”

“…wt + 1 = or * � (9) E +O'rel,i In Robert.S.Lee et.al [20], the data obtained from the relevant and irrelevant sets determined after user feedback are used to update the weights for the features by means of two-sided confidence intervals. Two-sided confidence intervals are used to make inferences on the differences between two sets proportions.…”

“…In the existing systems [20] [21] [22], feature weights lies completely in the range [-I, 1]. This range further divided into different intervals such as positive interval [0, 1], negative interval [-I, 0] and combination of both intervals.…”

Weighted Feature Gaussian Kernel SVM for Emotion Recognition

“…One effective method to solve this problem is to perform feature weighting based on the obtained feedback. Some methods [27, 28] divided the facial image into some uniform subregions and returned the subregion result for feature weighting. There is no restriction on each feature, which provides freedom on how the feature representations are structured.…”