[2Pm] Vision

[2Pm-034] Differences in action unit selectivity of convolutional neural networks trained with different facial expression databases

○Mikio Inagaki^1,2, Tatsuro Ito³, Takashi Shinozaki^1,4, Ichiro Fujita^1,2,5 (1. CiNet, NICT, Suita, Japan, 2. Grad Sch Front Biosci, Osaka Univ, Suita, Japan, 3. Sch Eng Sci, Osaka Univ, Toyonaka, Japan, 4. Fac Inform, Kindai Univ, Higashi-Osaka, Japan, 5. Res Org Sci Tech, Ritsumeikan Univ, Kusatsu, Japan)

Keywords:emotion, facial movement, transfer learning, supervised learning