Unsupervised manifold learning through reciprocal kNN graph and Connected Components for image retrieval tasks

Pedronette, Daniel Carlos Guimarães; Gonçalves, Filipe Marcel Fernandes; Guilherme, Ivan Rizzo

doi:10.1016/j.patcog.2017.05.009

Cited by 57 publications

(36 citation statements)

References 67 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Notice that our approaches present effectiveness results superior or comparable to the baselines in the majority of the scenarios. [23] et al [24] et al [25] et al [26] et al [27] [28] et al [29] et al [30] et al [31] et al [32] [34] et al [35] et al [28] et al [25] et al [18] et al [ [36] et al [37] et al [32] et al [30] et al [38] Figure 6 presents examples of visual results for three different queries on the UKBench dataset. Each row shows the results for the queries that were selected as part of the combination.…”

Section: Resultsmentioning

confidence: 99%

Unsupervised Selective Rank Fusion on Content-Based Image Retrieval

Valem

Pedronette

2019

Anais Estendidos Do XXXII Conference on Graphics, Patterns and Images (SIBRAPI Estendido 2019)

View full text Add to dashboard Cite

Mainly due to the evolution of technologies to store and share images, the growth of image collections have been remarkable for years. Therefore, developing effective methods to index and retrieve such extensive available visual information is indispensable. The CBIR (Content-Based Image Retrieval) systems are one of the main solutions for image retrieval tasks. These systems are mainly supported by the use of different visual descriptors and machine learning methods. Despite the relevant advances in the area, mainly driven by deep learning technologies, accurately computing the similarity between images remains a complex task in various scenarios due to the well known semantic gap problem. As distinct features produce complementary ranking results with different effectiveness performance, a promising solution consists in combining them. However, how to decide which visual features to combine is a very challenging task. This work proposes three novel methods for selecting and combining ranked lists by estimating their effectiveness in an unsupervised way. The approaches were evaluated in five different image collections and several descriptors, achieving results comparable or superior to the state-of-the-art in most of the evaluated scenarios.

show abstract

Section: Resultsmentioning

confidence: 99%

Unsupervised Selective Rank Fusion on Content-Based Image Retrieval

Valem

Pedronette

2019

Anais Estendidos Do XXXII Conference on Graphics, Patterns and Images (SIBRAPI Estendido 2019)

View full text Add to dashboard Cite

show abstract

“…A brief look at Algorithm 1 reveals a lot of potential for parallelism. For instance, there are three big loops whose bodies could be executed concurrently with almost no interference: (i) authority score loop (lines 7-24); (ii) distance computation loop (lines 25-32); and (iii) sorting loop (lines [33][34][35]. Notice, however, that these loops need to be executed in order: it is not possible to start the distance computation loop before completing the authority score computation loop.…”

Section: Parallelization Strategymentioning

confidence: 99%

“…Diffusion process [18,50] , graph-based learning methods [47] , and iterative re-ranking techniques [36,37] are some approaches proposed in the literature. The intrinsic structure of the datasets has been also exploited by other recent manifold learning methods [2,25,33,34] , yielding significant effectiveness gains in different scenarios.…”

Section: Introductionmentioning

confidence: 99%

An optimized unsupervised manifold learning algorithm for manycore architectures

Baldassin

Weng

Pedronette

et al. 2019

Information Sciences

View full text Add to dashboard Cite

Multimedia data, such as images and videos, has become very popular in people's daily life as a result of the widespread use of mobile devices. The ever-increasing amount of such data, along with the necessity for real-time retrieval, has lead to the development of new methods that can process them in a timely fashion with acceptable accuracy. In this paper, we study the performance of ReckNN, an unsupervised manifold learning algorithm based on the reciprocal neighbourhood and the authority of ranked lists. Most of the related work in this field do not fully investigate optimization strategies, an aspect that is becoming more important with the high availability of manycore machines. In order to address that issue, we fully investigate optimization opportunities in this article and make the following three main contributions. Firstly, we develop an efficient and scalable method for storing and accessing the distances between objects (e.g., video or image) based on dictionaries. Secondly, we employ memoization to speed up the computation of authority scores, leading to a significant performance gain even on single-core architectures. Lastly, we devise and implement several parallelization strategies and show that they are scalable on a 72-core Intel machine. The experimental results with MPEG-7, Corel5k and MediaEval benchmarks show that the optimized ReckNN delivers both efficiency and scalability, highlighting the importance of the proposed optimizations for manycore machines.

show abstract

“…The KNN algorithm is a widely used classification technique [39][40][41]. It is one of the simplest algorithms as it stocks all available cases and classifies the new ones by similarity measures.…”

Section: Classificationmentioning

confidence: 99%

An Assistive Technology for Braille Users to Support Mathematical Learning: A Semantic Retrieval System

2018

View full text Add to dashboard Cite

Mathematical learning from digital libraries and the web is a challenging problem for people with visual impairments and blindness. In this paper, we focus on developing the mathematical learning skills of braille users with a new assistive technology developed to retrieve semantically mathematical information from the web. This kind of research is still in the study phase. This paper presents an overview of assistive technologies for braille users followed by a description of the proposed system, which works in four phases. In the first phase, we translate a query math formula in braille into MathML code, and then we extract the structural and semantic meaning from the MathML expressions using multilevel presentation. In the classification phase, we choose a multilevel similarity measure based on K-Nearest Neighbors to evaluate the relevance between expressions. Finally, the query result is converted to braille math expressions. Experiments based on our database show that the new system provides more efficient results in responding to user queries.

show abstract

Unsupervised manifold learning through reciprocal kNN graph and Connected Components for image retrieval tasks

Cited by 57 publications

References 67 publications

Unsupervised Selective Rank Fusion on Content-Based Image Retrieval

Unsupervised Selective Rank Fusion on Content-Based Image Retrieval

An optimized unsupervised manifold learning algorithm for manycore architectures

An Assistive Technology for Braille Users to Support Mathematical Learning: A Semantic Retrieval System

Contact Info

Product

Resources

About