Upgrade to Pro — share decks privately, control downloads, hide ads and more …

惑星探査育英会・第十二回実習会

Sponsored · Your Podcast. Everywhere. Effortlessly. Share. Educate. Inspire. Entertain. You do you. We'll handle the rest.
Avatar for Sonochy Sonochy
March 09, 2020

 惑星探査育英会・第十二回実習会

Avatar for Sonochy

Sonochy

March 09, 2020
Tweet

Other Decks in Education

Transcript

  1. %FFQ-FBSOJOHͰ ը૾ࣝผ 2020/03/09 CPS 探査データ実習会@会津大学 The University of Aizu, Graduate

    School of Computer Science and Engineering. Graduate Department of Computer and Information Systems. ARC-Space
  2. Data Preparation Phase %BUBGMPX l4JNQMFJNBHF$MBTTJGJFSz B A Original data A&B

    Train data A&B Test data Divide Train and Test (into 8 : 2)
  3. Testing Phase Learning Phase %BUBGMPX l4JNQMFJNBHF$MBTTJGJFSz Learning Model (CNN) A&B

    Training data input Identification Model output input Output Image predictions A&B Training data Image_001 is A Image_002 is A Image_003 is B ….. Identification Model
  4. • サンプルコード Simple image ClassifierのCNNの構成は VGG16を参考にしている。 • 構成やパラメータによって識別率は変化する。 • 扱うデータに合わせて構成、

    パラメータを変更。 -FBSOJOH.PEFM $// ͷத਎ • Convolution (32,(3,3)) • Activation (relu) • Max_pooling (2,2) • Convolution (32,(3,3)) • Activation (relu) • Max_pooling (2,2) • Convolution (32,(3,3)) • Activation (relu) • Flatten • Dense (64) • Activation (relu) • Dense (class_num) • Activation (softmax) Learning Model (CNN)