RESUMEN
The production process of a wafer in the semiconductor industry consists of several phases such as a diffusion and associated defectivity test, parametric test, electrical wafer sort test, assembly and associated defectivity tests, final test, and burn-in. Among these, the fault detection phase is critical to maintain the low number and the impact of anomalies that eventually result in a yield loss. The understanding and discovery of the causes of yield detractors is a complex procedure of root-cause analysis. Many parameters are tracked for fault detection, including pressure, voltage, power, or valve status. In the majority of the cases, a fault is due to a combination of two or more parameters, whose values apparently stay within the designed and checked control limits. In this work, we propose an ensembled anomaly detector which combines together univariate and multivariate analyses of the fault detection tracked parameters. The ensemble is based on three proposed and compared balancing strategies. The experimental phase is conducted on two real datasets that have been gathered in the semiconductor industry and made publicly available. The experimental validation, also conducted to compare our proposal with other traditional anomaly detection techniques, is promising in detecting anomalies retaining high recall with a low number of false alarms.
Asunto(s)
Algoritmos , Semiconductores , DifusiónRESUMEN
Background: Breast shape is defined utilizing mainly qualitative assessment (full, flat, ptotic) or estimates, such as volume or distances between reference points, that cannot describe it reliably. Objectives: The authors quantitatively described breast shape with two parameters derived from a statistical methodology denominated by principal component analysis (PCA). Methods: The authors created a heterogeneous dataset of breast shapes acquired with a commercial infrared 3-dimensional scanner on which PCA was performed. The authors plotted on a Cartesian plane the two highest values of PCA for each breast (principal components 1 and 2). Testing of the methodology on a preoperative and posttreatment surgical case and test-retest was performed by two operators. Results: The first two principal components derived from PCA characterize the shape of the breast included in the dataset. The test-retest demonstrated that different operators obtain very similar values of PCA. The system is also able to identify major changes in the preoperative and posttreatment stages of a two-stage reconstruction. Even minor changes were correctly detected by the system. Conclusions: This methodology can reliably describe the shape of a breast. An expert operator and a newly trained operator can reach similar results in a test/re-testing validation. Once developed and after further validation, this methodology could be employed as a good tool for outcome evaluation, auditing, and benchmarking.