LithoROC: lithography hotspot detection with explicit ROC optimization. Ye, W., Lin, Y., Li, M., Liu, Q., & Pan, D. Z. In Proceedings of the 24th Asia and South Pacific Design Automation Conference on - ASPDAC '19, pages 292–298, Tokyo, Japan, 2019. ACM Press.
Paper doi abstract bibtex As modern integrated circuits scale up with escalating complexity of layout design patterns, lithography hotspot detection, a key stage of physical verification to ensure layout finishing and design closure, has raised a higher demand on its efficiency and accuracy. Among all the hotspot detection approaches, machine learning distinguishes itself for achieving high accuracy while maintaining low false alarms. However, due to the class imbalance problem, the conventional practice which uses the accuracy and false alarm metrics to evaluate different machine learning models is becoming less effective. In this work, we propose the use of the area under the ROC curve (AUC), which provides a more holistic measure for imbalanced datasets compared with the previous methods. To systematically handle class imbalance, we further propose the surrogate loss functions for direct AUC maximization as a substitute for the conventional cross-entropy loss. Experimental results demonstrate that the new surrogate loss functions are promising to outperform the cross-entropy loss when applied to the state-of-the-art neural network model for hotspot detection.
@inproceedings{ye_lithoroc_2019,
address = {Tokyo, Japan},
title = {{LithoROC}: lithography hotspot detection with explicit {ROC} optimization},
isbn = {978-1-4503-6007-4},
shorttitle = {{LithoROC}},
url = {http://dl.acm.org/citation.cfm?doid=3287624.3288746},
doi = {10.1145/3287624.3288746},
abstract = {As modern integrated circuits scale up with escalating complexity of layout design patterns, lithography hotspot detection, a key stage of physical verification to ensure layout finishing and design closure, has raised a higher demand on its efficiency and accuracy. Among all the hotspot detection approaches, machine learning distinguishes itself for achieving high accuracy while maintaining low false alarms. However, due to the class imbalance problem, the conventional practice which uses the accuracy and false alarm metrics to evaluate different machine learning models is becoming less effective. In this work, we propose the use of the area under the ROC curve (AUC), which provides a more holistic measure for imbalanced datasets compared with the previous methods. To systematically handle class imbalance, we further propose the surrogate loss functions for direct AUC maximization as a substitute for the conventional cross-entropy loss. Experimental results demonstrate that the new surrogate loss functions are promising to outperform the cross-entropy loss when applied to the state-of-the-art neural network model for hotspot detection.},
language = {en},
urldate = {2019-02-28},
booktitle = {Proceedings of the 24th {Asia} and {South} {Pacific} {Design} {Automation} {Conference} on - {ASPDAC} '19},
publisher = {ACM Press},
author = {Ye, Wei and Lin, Yibo and Li, Meng and Liu, Qiang and Pan, David Z.},
year = {2019},
keywords = {\#broken, Jab/\#ASPDAC},
pages = {292--298},
}
Downloads: 0
{"_id":"kPf8ETimnFi52LsZd","bibbaseid":"ye-lin-li-liu-pan-lithoroclithographyhotspotdetectionwithexplicitrocoptimization-2019","author_short":["Ye, W.","Lin, Y.","Li, M.","Liu, Q.","Pan, D. Z."],"bibdata":{"bibtype":"inproceedings","type":"inproceedings","address":"Tokyo, Japan","title":"LithoROC: lithography hotspot detection with explicit ROC optimization","isbn":"978-1-4503-6007-4","shorttitle":"LithoROC","url":"http://dl.acm.org/citation.cfm?doid=3287624.3288746","doi":"10.1145/3287624.3288746","abstract":"As modern integrated circuits scale up with escalating complexity of layout design patterns, lithography hotspot detection, a key stage of physical verification to ensure layout finishing and design closure, has raised a higher demand on its efficiency and accuracy. Among all the hotspot detection approaches, machine learning distinguishes itself for achieving high accuracy while maintaining low false alarms. However, due to the class imbalance problem, the conventional practice which uses the accuracy and false alarm metrics to evaluate different machine learning models is becoming less effective. In this work, we propose the use of the area under the ROC curve (AUC), which provides a more holistic measure for imbalanced datasets compared with the previous methods. To systematically handle class imbalance, we further propose the surrogate loss functions for direct AUC maximization as a substitute for the conventional cross-entropy loss. Experimental results demonstrate that the new surrogate loss functions are promising to outperform the cross-entropy loss when applied to the state-of-the-art neural network model for hotspot detection.","language":"en","urldate":"2019-02-28","booktitle":"Proceedings of the 24th Asia and South Pacific Design Automation Conference on - ASPDAC '19","publisher":"ACM Press","author":[{"propositions":[],"lastnames":["Ye"],"firstnames":["Wei"],"suffixes":[]},{"propositions":[],"lastnames":["Lin"],"firstnames":["Yibo"],"suffixes":[]},{"propositions":[],"lastnames":["Li"],"firstnames":["Meng"],"suffixes":[]},{"propositions":[],"lastnames":["Liu"],"firstnames":["Qiang"],"suffixes":[]},{"propositions":[],"lastnames":["Pan"],"firstnames":["David","Z."],"suffixes":[]}],"year":"2019","keywords":"#broken, Jab/#ASPDAC","pages":"292–298","bibtex":"@inproceedings{ye_lithoroc_2019,\n\taddress = {Tokyo, Japan},\n\ttitle = {{LithoROC}: lithography hotspot detection with explicit {ROC} optimization},\n\tisbn = {978-1-4503-6007-4},\n\tshorttitle = {{LithoROC}},\n\turl = {http://dl.acm.org/citation.cfm?doid=3287624.3288746},\n\tdoi = {10.1145/3287624.3288746},\n\tabstract = {As modern integrated circuits scale up with escalating complexity of layout design patterns, lithography hotspot detection, a key stage of physical verification to ensure layout finishing and design closure, has raised a higher demand on its efficiency and accuracy. Among all the hotspot detection approaches, machine learning distinguishes itself for achieving high accuracy while maintaining low false alarms. However, due to the class imbalance problem, the conventional practice which uses the accuracy and false alarm metrics to evaluate different machine learning models is becoming less effective. In this work, we propose the use of the area under the ROC curve (AUC), which provides a more holistic measure for imbalanced datasets compared with the previous methods. To systematically handle class imbalance, we further propose the surrogate loss functions for direct AUC maximization as a substitute for the conventional cross-entropy loss. Experimental results demonstrate that the new surrogate loss functions are promising to outperform the cross-entropy loss when applied to the state-of-the-art neural network model for hotspot detection.},\n\tlanguage = {en},\n\turldate = {2019-02-28},\n\tbooktitle = {Proceedings of the 24th {Asia} and {South} {Pacific} {Design} {Automation} {Conference} on - {ASPDAC} '19},\n\tpublisher = {ACM Press},\n\tauthor = {Ye, Wei and Lin, Yibo and Li, Meng and Liu, Qiang and Pan, David Z.},\n\tyear = {2019},\n\tkeywords = {\\#broken, Jab/\\#ASPDAC},\n\tpages = {292--298},\n}\n\n\n\n","author_short":["Ye, W.","Lin, Y.","Li, M.","Liu, Q.","Pan, D. Z."],"key":"ye_lithoroc_2019","id":"ye_lithoroc_2019","bibbaseid":"ye-lin-li-liu-pan-lithoroclithographyhotspotdetectionwithexplicitrocoptimization-2019","role":"author","urls":{"Paper":"http://dl.acm.org/citation.cfm?doid=3287624.3288746"},"keyword":["#broken","Jab/#ASPDAC"],"metadata":{"authorlinks":{}},"html":""},"bibtype":"inproceedings","biburl":"https://bibbase.org/zotero/bxt101","dataSources":["Wsv2bQ4jPuc7qme8R"],"keywords":["#broken","jab/#aspdac"],"search_terms":["lithoroc","lithography","hotspot","detection","explicit","roc","optimization","ye","lin","li","liu","pan"],"title":"LithoROC: lithography hotspot detection with explicit ROC optimization","year":2019}