vạn năng

  • Sự miêu tả :

Bộ dữ liệu Omniglot dành cho việc học một lần. Bộ dữ liệu này chứa 1623 ký tự viết tay khác nhau từ 50 bảng chữ cái khác nhau.

Tách ra Ví dụ
'small1' 2.720
'small2' 3.120
'test' 13.180
'train' 19.280
  • Cấu trúc tính năng :
FeaturesDict({
    'alphabet': ClassLabel(shape=(), dtype=int64, num_classes=50),
    'alphabet_char_id': int64,
    'image': Image(shape=(105, 105, 3), dtype=uint8),
    'label': ClassLabel(shape=(), dtype=int64, num_classes=1623),
})
  • Tài liệu tính năng :
Tính năng Lớp học Hình dạng Kiểu dữ liệu Sự miêu tả
Tính năngTừ điển
bảng chữ cái Nhãn lớp int64
alphabet_char_id Tensor int64
hình ảnh Hình ảnh (105, 105, 3) uint8
nhãn Nhãn lớp int64

Hình dung

  • Trích dẫn :
@article{lake2015human,
  title={Human-level concept learning through probabilistic program induction},
  author={Lake, Brenden M and Salakhutdinov, Ruslan and Tenenbaum, Joshua B},
  journal={Science},
  volume={350},
  number={6266},
  pages={1332--1338},
  year={2015},
  publisher={American Association for the Advancement of Science}
}