深層学習で切り拓くパーソナルロボットの未来 @東京大学 先端技術セミナー 工学最前線

1. 0 1Preferred Networks
2. l l Chainer CuPy OSS l 2
3. Preferred Networks (PFN) l 2014 3 l l l 3
4. 6 G $&"5&$
   +"1"/
   ˏນுϝοη ೥݄೔
6. × l l 6
8. CEATEC JAPAN 2018 (Oct 16–19, 2018) 8
13. 0 1 l l 2012 l He. et.al, Deep Residual Learning for Image Recognition. CVPR 2016.
    
15. 2 3 2 3 15
16. 2 3 16
20. 20 /
21. l l l l l l 21
22. l l (RGBD) l (RGB) l 4 (RGB) l l 300 22
23. 23 Sparse Dense
24. 2 3 24
28. Needs English subtitles
30. Needs English subtitles
36. 36 AR
37. PFDet l Google 2 l 170 1200 l NVIDIA Tesla V100 512 37 T. Akiba, et.al. PFDet: 2nd Place Solution to Open Images Challenge 2018 Object Detection Track. ECCV2018.
38. l l 38 K. Takahashi, J. Tan. Deep Visuo-Tactile Learning: Estimation of Tactile Properties from Images. ICRA2019.
39. l l 39 J. Hatori, et.al. Interactively Picking Real-World Objects with Unconstrained Spoken Language Instructions. ICRA2018.
41. 41 NVIDIA Tesla V100 (32GB) 512 Infiniband P
45. l 2 3 l 45
46. l G l l 46
47. l 2 l l l l 47
50. 50 55
51. l l l l 0 1 l 0 1 l l l 51
52. l l l 52
53. × l 0 1 l 0 1 l 53
54. 15 l l 2 3 l 54
55. 55 Photo by Kuha455405 / CC BY-SA 3.0
56. × l MD l MP3 1, 2 1 l × l × 9 01 56
57. l G l 57 250 8 3
58. 0 1 1 100 58 l l 0 1 U l l 0 1
59. l l l l l l 59
61. l l l × 61
63. 1 l l 4 l 63
64. PFI/PFN 64 OSS Blog TechTalk
65. × l l l 2 3 l 2 3 65
66. l 1 l G 2 66
67. l l l l 295 3 l 25 3 l l 67
68. l l l 68