Coping with recall and precision of soft error detectors

Abstract

International audienceMany methods are available to detect silent errors in high-performance computing (HPC) applications. Each method comes with a cost, a recall (fraction of all errors that are actually detected, i.e., false negatives), and a precision (fraction of true errors amongst all detected errors, i.e., false positives). The main contribution of this paper is to characterize the optimal computing pattern for an application: which detector(s) to use, how many detectors of each type to use, together with the length of the work segment that precedes each of them. We first prove that detectors with imperfect precisions offer limited usefulness. Then we focus on detectors with perfect precision , and we conduct a comprehensive complexity analysis of this optimization problem, showing NP-completeness and designing an FPTAS (Fully Polynomial-Time Approximation Scheme). On the practical side, we provide a greedy algorithm, whose performance is shown to be close to the optimal for a realistic set of evaluation scenarios. Extensive simulations illustrate the usefulness of detectors with false negatives, which are available at a lower cost than the guaranteed detectors

Extraction of mosaic regions through projection and filtering of features from image big data

“…Figure 10 shows the result of the actual mosaic blocks remaining after removing nonmosaic blocks from the candidate mosaic blocks through the application of geometrical features. In this paper, we used an accuracy measure defined as (17) and (18) to quantitatively evaluate the performance of the proposed mosaic detection algorithm [59][60][61][62][63]. In (17) and (18), N TP indicates the number of mosaic regions accurately detected, N FP represents the number of regions that are incorrectly detected as mosaic regions but not mosaic regions, and N FN denotes the number of mosaic regions that are not detected.…”

Efficient Image Recognition and Retrieval on IoT-Assisted Energy-Constrained Platforms From Big Data Repositories

“…It is more challenging because the images are taken with different illumination changes, indoor, and outdoor environments. The proposed system is evaluated using precision, recall score [41] and MAP score [42]. Our system is also assessed and compared with different handcrafted feature based methods including the local binary pattern (LBP) [43], HOG [44], SURF [11], and color texture fused features [45] A well-known information retrieval evaluation method "precision and recall" is used for the performance assessment of the proposed system.…”

Extract hidden patterns in students' academic information to improve the curriculum by using data mining

“…where P is precision and R is recall. Therefore, the F criterion ( 2) is considered as a criterion to compare results [18]. RMSE is utilized for the prediction error.…”