Online learning algorithms for nonlinear regression of time - varying systems
In industry, there often exists a need from manufacturers for a model that accurately predicts machine yield and allows further fine-tuning of input settings to improve quality without physical testing; saving on cost and time. And this can be achieved by machine learning, since machine learning is...
Saved in:
| 主要作者: | |
|---|---|
| 其他作者: | |
| 格式: | Final Year Project |
| 語言: | English |
| 出版: |
2015
|
| 主題: | |
| 在線閱讀: | http://hdl.handle.net/10356/63910 |
| 標簽: |
添加標簽
沒有標簽, 成為第一個標記此記錄!
|
| 總結: | In industry, there often exists a need from manufacturers for a model that accurately predicts machine yield and allows further fine-tuning of input settings to improve quality without physical testing; saving on cost and time. And this can be achieved by machine learning, since machine learning is to build prediction model based on historical data collected. This project seeks to investigate the optimal method for modelling an Inkjet Page Yield Testing System using real production data, by analyzing and evaluating the models in terms of training speed and accuracy. In this project, three typical machine learning methods are chosen, 2 of them online--Extreme Learning Machine, Quantized Kernel Least Mean Square, 1 of them off-line--Gaussian Process Regression, to do the simulation. |
|---|