Evidential framework for Error Correcting Output Code classification

Lachaize, Marie; Hégarat‐Mascle, Sylvie Le; Aldea, Emanuel; Maitrot, Aude; Reynaud, Roger

doi:10.1016/j.engappai.2018.04.019

Cited by 27 publications

(17 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Such an 210 approach is suitable specially when the number of samples if small and when there is no overlapping between the scores of the two considered classes. Otherwise, [45] shows that, in such difficult types of applications, it is hard for SVM to find a very large margin between the two classes and there can be a consistent overlap between samples with different labels for the same score. However, since the number of samples per 215 score would be high, we would paradoxically not assign a high value of imprecision to them.…”

Section: Bba Definition Based On Calibrated Scoresmentioning

confidence: 99%

“…Popular strategies consist in querying the instance whose predicted output is the least confident or with maximum entropy, but in the context of SVM classification the prevailing method is to select the samples which are closer to the separation hyperplane margin [20,21]. More recently, DUAL [22] and QUIRE [23] methods have 45 been proposed. The former is based on density weighted uncertainty sampling while the latter aims at selecting both informative and representative examples on the basis of a prediction of the uncertainty.…”

Section: Introductionmentioning

confidence: 99%

“…In the context of high-density crowd pedestrian detection, in [39] we proposed a robust fusion strategy based on the Belief Functions (BF) framework [40,41,42]. The 110 evidential framework [43,44,36,45] is indeed able to naturally model the concept of imprecision, that in our case can arise in two different and complementary ways: in the derivation of posterior probability values from SVM decision scores, and later, from the spatial layout of the detections in the output image space. The proposed mass allocation has been shown to be robust to possible imprecision of the calibration functions, while 115 at the same time taking into account the information coming from neighboring pixels in the image space.…”

mentioning

confidence: 99%

See 2 more Smart Citations

Evidential query-by-committee active learning for pedestrian detection in high-density crowds

Vandoni

Aldea

Hégarat‐Mascle

2019

International Journal of Approximate Reasoning

Self Cite

View full text Add to dashboard Cite

Mascle. Evidential query-by-committee active learning for pedestrian detection in high-density crowds. International Journal of Approximate Reasoning, Elsevier, In press, AbstractThe automatic detection of pedestrians in dense crowds has become recently a very active topic of research due to the implications for public safety, and also due to the increased frequency of large scale social events. The detection task is complicated by multiple factors such as strong occlusions, high homogeneity, small target size, etc., and different types of detectors are able to provide complementary interpretations of the input data, with varying individual levels of performance. Our first contribution consists in outlining a fusion strategy under the form of an ensemble method, which models the imprecision arising from each of the detectors, both in the calibration and in the spatial domains in an evidential framework. Then, we propose a sample selection for augmenting the training set used jointly by the committee of classifiers, based on evidential disagreement measures among the base members in a Query-by-Committee context. The results show that the proposed fusion algorithm is effective in exploiting the strengths of the individual classifiers, as well as in augmenting the training set with informative samples which allow the resulting detector to enhance its performance.

show abstract

Section: Bba Definition Based On Calibrated Scoresmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

mentioning

confidence: 99%

See 1 more Smart Citation

Evidential query-by-committee active learning for pedestrian detection in high-density crowds

Vandoni

Aldea

Hégarat‐Mascle

2019

International Journal of Approximate Reasoning

Self Cite

View full text Add to dashboard Cite

show abstract

“…In order to handle both the uncertainty and the imprecision during the classification, we recently proposed an evidential ECOC approach, [12]. Here we are interested in the ECOC decoding part of this work.…”

Section: Experimental Settingsmentioning

confidence: 99%

“…In brief, this part works as follows. An ECOC can be viewed as a set of l dichotomizers (i.e., binary classifiers, SVM in our case) for which the two considered hypotheses are non-overlapping subsets of Ω. Denoting these two hypotheses A i,0 and A i,1 for the i th SVM, its output score for a given sample p is converted (see [12] for details) into an elementary bba m b i having A i,0 , A i,1 and A i,0 ∪A i,1 as focal elements. For non-dense ECOC, a deconditioning on A i,0 ∪ A i,1 allows us to model the absence of information on A i,0 ∪ A i,1 provided by this i th dichotomizer.…”

Section: Experimental Settingsmentioning

confidence: 99%

Evidential split-and-merge: Application to object-based image analysis

Lachaize

Hégarat‐Mascle

Aldea

et al. 2018

International Journal of Approximate Reasoning

Self Cite

View full text Add to dashboard Cite

This paper addresses the difficult problem of segmenting objects in a scene and simultaneously estimating their material class. Focusing on the case where, individually, no dataset can achieve such a task, multiple sensor datasets are considered, including some images for retrieving the spatial information. The proposed approach is based on mutual validation between class decision (using the most relevant dataset) and segmentation (derived from image data). The main originality relies in the ability to make these two modules (classification and segmentation) interactive. Specifically, our application focuses on object-level material labeling using classic RGB images, laser profilometer images and a NIR spectral sensor. Starting from a superpixel segmentation, the relevant data are introduced as constraints modifying the initial segmentation in a splitand-merge process, which interacts with the material labeling process. In this work, we use the belief function framework to model the information extracted from each kind of data and to transfer it from one processing module to another. In particular we show the relevance of evidential conflict measure to drive the split process and to control the merge one. Experiments have been performed on actual scenes with stacked objects and difficult cases of material such as transparent polymers. They allow us to assess the performance of the proposed approach both in terms of material labeling and object segmentation as well as to illustrate some borderline cases.

show abstract

A new re-encoding ECOC using reject option

2020

View full text Add to dashboard Cite

When training base classifier by ternary Error Correcting Output Codes (ECOC), it is well know that some classes are ignored. On this account, a non-competent classifier emerges when it classify an instance whose real label does not belong to the meta-subclasses. Meanwhile, the classic ECOC dichotomizers can only produce binary outputs and have no capability of rejection for classification. To overcome the non-competence problem and better model the multi-class problem for reducing the classification cost, we embed reject option to ECOC and present a new variant of ECOC algorithm called as Reject-Option-based Re-encoding ECOC (ROECOC). The cost-sensitive classification model and cost-loss function based on Receiver Operating Characteristic (ROC) curve are built respectively. The optimal reject threshold values are obtained by combing the condition to be met for minimizing the loss function and the ROC convex hull. In so doing, reject option (t1, t2) provides a three-symbol output to make dichotomizers more competent and ROECOC more universal and practical for cost-sensitive classification issue. Experimental results on two kinds of datasets show that our scheme with low-degree freedom of initialized ECOC can effectively enhance accuracy and reduce cost.

show abstract

Evidential framework for Error Correcting Output Code classification

Cited by 27 publications

References 32 publications

Evidential query-by-committee active learning for pedestrian detection in high-density crowds

Evidential query-by-committee active learning for pedestrian detection in high-density crowds

Evidential split-and-merge: Application to object-based image analysis

A new re-encoding ECOC using reject option

Contact Info

Product

Resources

About