Deep HTM for GPU - Differentiable Scalar Spatial Pooler / 미분가능한 스칼라 공간풀러
The whole program is written in C++ with OpenGL(GLSL Compute Shader). The internal process is pretty much the same as the paper Numenta wrote describes. I've set up the network be an autoencoder. The weird thing is that the weight matrix of SP is pretty dense as opposed to the permanence matrix of the vanilla SP being almost as sparse as the input. And That's what I expected to happen to it as w..
연구
2019. 10. 9. 22:07
공지사항
최근에 올라온 글
최근에 달린 댓글
- Total
- Today
- Yesterday
링크
TAG
- spatial pooler
- 공간풀러
- 인공신경망
- 딥러닝
- GPGPU
- 공간 풀러
- hierarchical temporal memory
- 계층형 시간적 메모리
- Compute Shader
- htm
- opengl
- 역전파
- 신경망
- GLSL
- mnist
일 | 월 | 화 | 수 | 목 | 금 | 토 |
---|---|---|---|---|---|---|
1 | 2 | 3 | 4 | |||
5 | 6 | 7 | 8 | 9 | 10 | 11 |
12 | 13 | 14 | 15 | 16 | 17 | 18 |
19 | 20 | 21 | 22 | 23 | 24 | 25 |
26 | 27 | 28 | 29 | 30 | 31 |
글 보관함