Direct maximization of average precision by hill-climbing, with a comparison to a maximum entropy approach

Morgan, W. T. W.; Greiff, Warren R.; Henderson, J.H.

doi:10.3115/1613984.1614008

Cited by 21 publications

(13 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In this respect, good retrieval performance requires that the model do well with many different kinds of "word" queries while annotation requires one to do well at finding as many keywords as possible for images. This issue has also been observed in text retrieval where it has been shown that the likelihood surface is unlikely to correlate with the retrieval metric surface [21,19].…”

Section: Introductionmentioning

confidence: 85%

A discrete direct retrieval model for image and video retrieval

Feng

Manmatha

2008

Proceedings of the 2008 International Conference on Content-Based Image and Video Retrieval

View full text Add to dashboard Cite

This paper proposes a formal framework for image and video retrieval using discrete Markov random fields (MRF). The training dataset consists of images with keywords (regions are not labeled). The model is built using a discrete vocabulary of vector quantized region or point features generated from the training images. Since performance is dependent on the size of the vocabulary, a large vocabulary of a couple of million visterms is used. Such large vocabularies cannot be generated by conventional clustering algorithms so hierarchical k-means is used to generate it. Unlike many previous techniques, our MRF based model doesn't require an explicit annotation step for retrieval. The model directly ranks all test images according to the posterior probability of an image given a query. Traditionally, most models are trained by maximizing likelihood -instead this model is trained by maximizing average precision. Image and video retrieval experiments are performed on two standard datasets (a Corel dataset and a TRECVID3 dataset) which consist of 4,500 images and about 44,100 keyframes respectively. The results show that based on a large visual vocabulary the model runs extremely fast on even very large datasets while having comparable retrieval performance to the best performing (continuous feature) models.

show abstract

Section: Introductionmentioning

confidence: 85%

A discrete direct retrieval model for image and video retrieval

Feng

Manmatha

2008

Proceedings of the 2008 International Conference on Content-Based Image and Video Retrieval

View full text Add to dashboard Cite

show abstract

“…However, this has been shown experimentally to be invalid and it can also be shown theoretically to be invalid, as well. This is known as metric divergence [22].…”

Section: Parameter Estimationmentioning

confidence: 99%

Linear feature-based models for information retrieval

2007

View full text Add to dashboard Cite

Abstract. There have been a number of linear, feature-based models proposed by the information retrieval community recently. Although each model is presented differently, they all share a common underlying framework. In this paper we explore and discuss the theoretical issues of this framework, including a novel look at the parameter space. We then detail supervised training algorithms that directly maximize the evaluation metric under consideration, such as mean average precision. We present results that show training models in this way can lead to significantly better test set performance compared to other training methods that do not directly maximize the metric. Finally, we show that linear feature-based models can consistently and significantly outperform current state of the art retrieval models with the correct choice of features.

show abstract

“…Feng and Manmatha [9] showed that this approach worked for image retrieval and we also maximised average precision. We followed a hill-climbing mean average precision optimisation as explained in [25].…”

Section: Model Trainingmentioning

confidence: 99%

Image retrieval using Markov Random Fields and global image features

Llorente

Manmatha

Rüger

2010

Proceedings of the ACM International Conference on Image and Video Retrieval

View full text Add to dashboard Cite

In this paper, we propose a direct image retrieval framework based on Markov Random Fields (MRFs) that exploits the semantic context dependencies of the image. The novelty of our approach lies in the use of different kernels in our non-parametric density estimation together with the utilisation of configurations that explore semantic relationships among concepts at the same time as low-level features, instead of just focusing on correlation between image features like in previous formulations. Hence, we introduce several configurations and study which one achieve the best performance. Results are presented for two datasets, the usual benchmark Corel 5k and the collection proposed by the 2009 edition of the ImageCLEF campaign. We observe that, using MRFs, performance increases significantly depending on the kernel used in the density estimation for the two datasets. With respect to the the language model, best results are obtained for the configuration that exploits dependencies between words together with dependencies between words and visual features. For the Corel 5k dataset, our best result corresponds to a mean average precision of 0.32, which compares favourably with the highest value ever obtained, 0.35, achieved by Makadia et al. [22] albeit with different features. For the ImageCLEF09 collection, we obtained 0.32, as mean average precision.

show abstract

Direct maximization of average precision by hill-climbing, with a comparison to a maximum entropy approach

Cited by 21 publications

References 7 publications

A discrete direct retrieval model for image and video retrieval

A discrete direct retrieval model for image and video retrieval

Linear feature-based models for information retrieval

Image retrieval using Markov Random Fields and global image features

Contact Info

Product

Resources

About