-
公开(公告)号：KR1020130134901A
公开(公告)日：2013-12-10
申请号：KR1020120058772
申请日：2012-05-31
Applicant: 서울대학교산학협력단
IPC: G06F19/00
Abstract: The present invention relates to a method for manufacturing a nonlinear activity prediction module of an unknown natural sample and a method for predicting the biological activity of the unknown natural sample. The biological activity of the unknown natural sample is effectively evaluated by the biological activity prediction model of the unknown natural sample and the method for predicting the biological activity of the unknown natural sample using the same according to one embodiment.
Abstract translation: 本发明涉及一种用于制造未知天然样品的非线性活动预测模块的方法和用于预测未知天然样品的生物活性的方法。 未知天然样品的生物活性通过未知天然样品的生物活性预测模型和根据一个实施方案使用该未知天然样品的未知天然样品的生物学活性的方法有效地进行评估。
-
公开(公告)号：KR102199285B1
公开(公告)日：2021-01-06
申请号：KR1020180150181
申请日：2018-11-28
Applicant: 서울대학교산학협력단
Abstract: 딥러닝신경망압축방법은, 상기딥러닝신경망을구성하는적어도하나의텐서(tensor)를분해하되, 동일한입력에대해서원래의텐서와분해된텐서의출력이가장근사해지도록상기적어도하나의텐서를분해하는단계및 상기원래의텐서및 분해된텐서모두에임의의노이즈를입력으로주고, 이들의출력을통해상기원래의텐서및 분해된텐서를학습시키는단계를포함한다.
-
Search Results
2
records
(took 0.015 seconds)
