Catálogo de publicaciones - libros

In the semiconductor manufacturing, yield enhancement is an important issue. It is ideal to prevent all failures. However, when a failure occurs, it is important to quickly specify the cause stage and take countermeasures. Reviewing wafer level and composite lot level yield patterns has always been an effective way of identifying yield inhibitors and driving process improvement. This process is very time consuming and as such generally occurs only when the overall yield of a device has dropped significantly enough to warrant investigation. The automatic method of failure pattern extraction from fail bit map provides reduced time to analysis and facilitates yield enhancement. The automatic method of failure pattern extraction from fail bit map provides reduced time to analysis and facilitates yield enhancement. This paper describes the techniques to automatically recognize and classifies a failure pattern using a fail bit map, a new simple schema which facilitates the failure analysis.

This research developed the CBD(Component Based Development) logic frameworks for the Integration of components and a tool which supports this. The Integration of components becomes necessary during the process of reusing or assembling components, and this is because the interface of the component is, in many cases, different than the component the developer wishes to assemble. After the iterations of specification modification and verification in terms of knowledge acquisition activities, Logic frameworks are correctly formed. Occasionally, additional attributes may need to be defined in accordance to new requirements. Consequently, the process for component Integration is crucial for the reuse and assembly of components. In order to support the Integration of components, this research proposes an Integration technique dependent upon binary component Integration techniques and Integration components. In addition, a support tool was developed to support an effective Integration process.

The EDS wafer test yield is the most important criteria to evaluate FAB’s productivity, so the manufacturing operation’s main purpose is to secure new product yield early and maintaining the yield of mass-produced products high. Defining a failed characteristic that’s compatible to the device and classifying wafers depending on failure type helps tasks searching for error from FAB become automated. This would be more efficient then existing failed analysis operations and strive to become the basis for improvement in yield and quality. For this method, this research is trying to use a high speed recognition algorithm called SVM (support vector machine) that will define wafer’s failed type and automatically classify each one.

An organization can optimize a project and strengthen control of it and thus, accomplish its objectives by determining its project capability through the proposed models, by planning the most suitable project to its vision. The approach combines two methods. One looks at an organization’s measurement framework in goal-oriented fashion and the other looks at it in vision driven. The goal-oriented method was applied to improve the quality and delivery measurement from the point of view. It identified several new metrics, and also contributed to better understanding the collected data of user requirement. The vision-driven method was used to gain new insights into the existing PCM(Performance Calculation Model) data.This paper gives the case study and its results to qualitatively compare our approach against current ad hoc practices used to improve existing Opportunity Tree frameworks(OTF).

Estimating animal’s genetic merit (or ) plays a major role in the Manchego sheep selection scheme (ESROM), started fifteen years ago with the goal of improving Manchego sheep production figures. In the ESROM scheme the breeding value is estimated each semester by using BLUP animal model. In this paper we study the use of data mining techniques to deal with breeding value classification. The purpose of the paper is not to replace the use of BLUP in the ESROM, on the contrary, we intend to learn in a supervised way from the results produced by BLUP, and to use the learned models to provide preliminary information about the breeding value of an animal. We use standard classification techniques combined with feature subset selection in order to identify good (subsets of) predictors. We also show that the classifiers accuracy can be considerably improved by attribute construction.

This paper presents a new learning method for automatic acquisition of translation knowledge from parallel corpora. We apply this learning method to automatic extraction of bilingual word pairs from parallel corpora. In general, similarity measures are used to extract bilingual word pairs from parallel corpora. However, similarity measures are insufficient because of the sparse data problem. The essence of our learning method is this presumption: in local parts of bilingual sentence pairs, the equivalents of words that adjoin the source language words of bilingual word pairs also adjoin the target language words of bilingual word pairs. Such adjacent information is acquired automatically in our method. We applied our method to systems based on various similarity measures, thereby confirming the effectiveness of our method.