Zur Webseite der Uni Stuttgart

Data Mining and Machine Learning Approaches for Semiconductor Test and Diagnosis

 

 

Overview

New manufacturing issues at the nanoscale level lead to process variations, increased sensitivity to environmental conditions and degradation. These emerging challenges threaten both system quality and reliability, and make the exact behavior of semiconductor technology exceedingly hard to predict. Consequently, the test and diagnosis of complex VLSI systems need to evolve in order to account for performance uncertainty.


In order to fulfill this goal, semiconductor measurements may be analyzed, which characterize semiconductor performance during manufacturing and system-level tests. The extracted performance characteristics can used to improve a wide array of tasks in the semiconductor industry like, for example, analog diagnosis, yield optimization, adaptive test, system-level diagnosis, etc.

This challenge is a natural fit for data mining and machine learning approaches, which are best suited for problems where there is no fixed set of rules that can be efficiently specified in an algorithm. Instead, these techniques analyze different characteristics in the data itself and try to infer relationships among them.

This seminar deals with the most useful tools for data mining and machine learning for semiconductor test and diagnosis. We will discuss the principles and application of various techniques, such as regression, classification (support vector machines and neural networks), clustering and outlier detection.

 

Staff

 

Material

     

    Schedule

    • The seminar will take place on Wednesdays, from 09:45 to 11:15, in 0.363.
    • The first meeting will take place on 09.04.14. It is mandatory for participation in this seminar.

    Week

    Date

    Time

    Room

    Topic

    Super-
    visor

    Material

    15

    09.04

    09:45 - 11:15

    0.363

    Introduction, Rules and Topics

    slides

    16

    23.04

    09:45 - 11:15

    0.363

    Introduction to Machine Learning

    AC

     

    19

    07.05

    09:45 - 11:15

    0.363

    Signature Generation for Signature Testing of Analog Components using Genetic Algorithms

    RB

    report

    Oscillation- and Loopback-based Testing of Mixed-Signal Circuit using Adaptive Regression

    RB

    report

    20

    14.05

    09:45 - 11:15

    0.363

    Diagnosis of Defects in Analog Circuits using Neural Networks

    RB

    report

    Dealing with Marginal Devices and Outliers in Analog Test

    AC

    report

    21

    21.05

    09:45 - 11:15

    0.363

    Dimensionality Reduction for Test Optimization

    AC

    report

    22

    28.05

    09:45 - 11:15

    0.363

    No seminar session

    23

    04.06

    09:45 - 11:15

    0.363

    No seminar session

    24

    11.06

    09:45 - 11:15

    0.363

    No seminar session

    25

    18.06

    09:45 - 11:15

    0.363

    No seminar session

    26

    25.06

    09:45 - 11:15

    0.363

    Board-Level Functional Test

    DU

    report

    27

    02.07

    09:45 - 11:15

    0.363

    Post-Silicon Timing Validation

    LRG

    report

    Analysis of Customer Field Returns

    ES

    report

    Akzeptieren

    Diese Webseite verwendet Cookies. Durch die Nutzung dieser Webseite erklären Sie sich damit einverstanden, dass Cookies gesetzt werden. Mehr erfahren, zum Datenschutz