Search research articles

ABOUT JoVE

Overview Leadership Blog JoVE Help Center

AUTHORS

Publishing Process Editorial Board Scope & Policies Peer Review FAQ Submit

LIBRARIANS

Testimonials Subscriptions Access Resources Library Advisory Board FAQ

RESEARCH

JoVE Journal Methods Collections JoVE Encyclopedia of Experiments Archive

EDUCATION

JoVE Core JoVE Business JoVE Science Education JoVE Lab Manual Faculty Resource Center Faculty Site

Terms & Conditions of Use

Search research articles

Related Experiment Video

Updated: Jun 21, 2026

Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness

Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness

Published on: December 6, 2024

Towards optimal indexing for relevance feedback in large image databases +.

Sharadh Ramaswamy¹, Kenneth Rose

¹Signal Compression Lab, Electrical and Computer Engineering, University of California, Santa Barbara, CA 93106, USA. rsharadh@ece.ucsb.edu

IEEE Transactions on Image Processing : a Publication of the IEEE Signal Processing Society

|August 5, 2009

Summary

We developed a fast, clustering-based indexing method for image retrieval that uses user relevance feedback. This technique efficiently adapts to varying Mahalanobis distances, improving search performance significantly.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by

Same author

Self-Reported Interpersonal Safety Concerns During Outdoor, Daytime Physical Activities by Sex and Race/Ethnicity, SummerStyles - United States, 2022.

American journal of health promotion : AJHP·2026

Same author

Emergency Department Visits for Pedestrians Injured in Motor Vehicle Traffic Crashes - United States, January 2021-December 2023.

MMWR. Morbidity and mortality weekly report·2024

Same author

Traffic as a barrier to walking safely in the United States: Perceived reasons and potential mitigation strategies.

Preventive medicine reports·2022

Same author

A Geodesic Translation Model for Spherical Video Compression.

IEEE transactions on image processing : a publication of the IEEE Signal Processing Society·2022

Same author

Effective Prediction Modes Design for Adaptive Compression With Application in Video Coding.

IEEE transactions on image processing : a publication of the IEEE Signal Processing Society·2021

Same author

The Chesapeake Bay Program Modeling System: Overview and Recommendations for Future Development.

Ecological modelling·2021

Area of Science:

Computer Science
Information Retrieval
Machine Learning

Background:

Content-based image retrieval (CBIR) systems require efficient methods to process user relevance feedback.
Existing indexing techniques may not optimally adapt to dynamic distance metrics like the Mahalanobis distance.

Purpose of the Study:

To propose a novel, fast, clustering-based indexing technique for exact nearest-neighbor search in image databases.
To enhance the efficiency of leveraging user relevance feedback in CBIR.

Main Methods:

Developed a clustering-based indexing approach for exact nearest-neighbor search.
Derived a property of point-to-hyperplane Mahalanobis distance for efficient distance recalculation.
Utilized this property to recalculate query-cluster distance bounds and prune non-competitive clusters.

Related Experiment Videos

Last Updated: Jun 21, 2026

Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness

Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness

Published on: December 6, 2024

Implemented a search strategy that retrieves clusters in increasing order of distance.

Main Results:

The proposed technique efficiently adapts to a varying Mahalanobis distance weight matrix.
Demonstrated effective pruning of irrelevant clusters during the search process.
Achieved considerable performance gains compared to an existing VA-File indexing variant for relevance feedback.

Conclusions:

The proposed clustering-based indexing technique offers a significant improvement for relevance feedback in CBIR.
Efficient adaptation to Mahalanobis distance and effective cluster pruning contribute to substantial performance gains.