-
Notifications
You must be signed in to change notification settings - Fork 0
Binding parameters to make CNN rotationally invariant
License
travers-rhodes/ricnn
Folders and files
| Name | Name | Last commit message | Last commit date | |
|---|---|---|---|---|
Repository files navigation
This model was trained very late monday night, early Tuesday morning, running code found at
commit d68beec344b6fd17e0d05b161948f064ea416bd7
Author: Travers Rhodes <traversr@andrew.cmu.edu>
Date: Mon Dec 4 12:09:49 2017 -0500
10,100,5, trained overnight
Its output is shown below
(tensorflow) coral@peppercmu:~/travers/ricnn/code$ python main.py
Extracting MNIST_data/train-images-idx3-ubyte.gz
Extracting MNIST_data/train-labels-idx1-ubyte.gz
Extracting MNIST_data/t10k-images-idx3-ubyte.gz
Extracting MNIST_data/t10k-labels-idx1-ubyte.gz
the size of the last conv layer is (?, 46, 46, 100)
the shape of our layer after maxpool (?, 1, 1, 100)
yconv has shape (?, 10)
2017-12-04 00:16:16.452808: I tensorflow/core/platform/cpu_feature_guard.cc:137] Your CPU supports instructions that this TensorFlow binary was not compiled to use: SSE4.1 SSE4.2 AVX AVX2 FMA
2017-12-04 00:16:16.536341: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:892] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2017-12-04 00:16:16.536702: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1030] Found device 0 with properties:
name: GeForce GTX 1080 Ti major: 6 minor: 1 memoryClockRate(GHz): 1.683
pciBusID: 0000:01:00.0
totalMemory: 10.91GiB freeMemory: 10.04GiB
2017-12-04 00:16:16.624406: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:892] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2017-12-04 00:16:16.624721: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1030] Found device 1 with properties:
name: GeForce GTX 1080 Ti major: 6 minor: 1 memoryClockRate(GHz): 1.683
pciBusID: 0000:02:00.0
totalMemory: 10.91GiB freeMemory: 10.75GiB
2017-12-04 00:16:16.625229: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1045] Device peer to peer matrix
2017-12-04 00:16:16.625280: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1051] DMA: 0 1
2017-12-04 00:16:16.625285: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1061] 0: Y Y
2017-12-04 00:16:16.625301: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1061] 1: Y Y
2017-12-04 00:16:16.625306: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1120] Creating TensorFlow device (/device:GPU:0) -> (device: 0, name: GeForce GTX 1080 Ti, pci bus id: 0000:01:00.0, compute capability: 6.1)
2017-12-04 00:16:16.625325: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1120] Creating TensorFlow device (/device:GPU:1) -> (device: 1, name: GeForce GTX 1080 Ti, pci bus id: 0000:02:00.0, compute capability: 6.1)
checking confusion
[[ 0 0 0 0 0 0 0 0 0 0]
[ 0 0 0 0 0 0 0 0 0 0]
[ 0 0 0 0 0 0 0 0 0 0]
[ 0 0 0 0 0 0 0 0 0 0]
[ 0 0 0 0 0 0 0 0 0 0]
[ 0 0 0 0 0 0 0 0 0 0]
[479 563 488 493 535 434 501 550 462 495]
[ 0 0 0 0 0 0 0 0 0 0]
[ 0 0 0 0 0 0 0 0 0 0]
[ 0 0 0 0 0 0 0 0 0 0]]
checking accuracy...
(0, 0.1002, 0.1002)
Model saved in file: /tmp/model.ckpt
checking accuracy...
(10000, 0.85460000000000003, 0.85460000000000003)
Model saved in file: /tmp/model.ckpt
checking confusion
[[477 0 0 2 0 0 1 0 2 8]
[ 0 557 3 2 2 0 0 5 0 1]
[ 1 2 374 11 14 26 10 14 0 4]
[ 0 0 10 465 0 17 1 1 7 2]
[ 0 4 32 2 485 1 9 44 3 31]
[ 0 0 26 2 4 363 2 0 1 11]
[ 0 0 10 1 4 8 440 8 1 126]
[ 1 0 16 3 20 2 12 476 0 1]
[ 0 0 16 4 4 2 2 0 447 4]
[ 0 0 1 1 2 15 24 2 1 307]]
checking accuracy...
(20000, 0.91059999999999997, 0.87819999999999998)
Model saved in file: /tmp/model.ckpt
checking accuracy...
(30000, 0.92179999999999995, 0.87619999999999998)
Model saved in file: /tmp/model.ckpt
checking confusion
[[471 0 0 0 0 0 1 0 1 6]
[ 0 550 1 0 1 0 0 1 1 0]
[ 0 3 356 5 15 9 7 9 2 1]
[ 1 0 11 472 3 10 2 1 7 0]
[ 0 1 15 3 453 1 3 18 4 12]
[ 0 0 42 3 4 346 2 2 1 4]
[ 1 3 13 2 3 7 431 11 3 155]
[ 4 6 35 4 43 6 12 504 1 0]
[ 0 0 8 1 1 2 1 0 434 1]
[ 2 0 7 3 12 53 42 4 8 316]]
checking accuracy...
(40000, 0.9224, 0.86660000000000004)
Model saved in file: /tmp/model.ckpt
checking accuracy...
(50000, 0.93440000000000001, 0.88419999999999999)
Model saved in file: /tmp/model.ckpt
checking confusion
[[464 0 0 0 0 0 1 0 0 2]
[ 0 551 1 0 0 0 0 1 0 0]
[ 2 3 390 6 18 14 3 8 3 3]
[ 0 0 8 463 1 4 2 3 8 1]
[ 0 3 16 0 486 1 3 37 2 10]
[ 1 1 32 12 5 389 5 5 5 10]
[ 3 0 17 2 3 0 432 20 1 124]
[ 5 5 7 4 9 1 6 475 0 0]
[ 1 0 5 2 0 0 1 0 435 2]
[ 3 0 12 4 13 25 48 1 8 343]]
checking accuracy...
(60000, 0.92459999999999998, 0.88560000000000005)
Model saved in file: /tmp/model.ckpt
checking accuracy...
(70000, 0.92800000000000005, 0.89459999999999995)
Model saved in file: /tmp/model.ckpt
checking confusion
[[467 0 0 0 0 0 0 0 0 3]
[ 0 554 2 0 0 0 0 2 0 0]
[ 1 1 368 8 8 9 3 6 0 1]
[ 0 0 6 473 0 4 2 2 2 0]
[ 0 5 24 0 503 1 5 27 1 3]
[ 0 0 47 5 6 394 4 5 2 8]
[ 0 0 8 0 0 0 392 16 1 85]
[ 4 3 12 2 8 1 7 489 0 0]
[ 1 0 10 3 0 1 3 0 451 3]
[ 6 0 11 2 10 24 85 3 5 392]]
checking accuracy...
(80000, 0.93540000000000001, 0.89659999999999995)
Model saved in file: /tmp/model.ckpt
checking accuracy...
(90000, 0.93540000000000001, 0.88319999999999999)
Model saved in file: /tmp/model.ckpt
checking confusion
[[473 0 1 0 0 0 1 0 1 6]
[ 0 556 4 1 0 1 0 6 1 0]
[ 1 1 411 5 17 18 5 4 3 3]
[ 0 0 7 485 7 16 2 4 7 2]
[ 0 3 8 0 475 0 3 12 0 6]
[ 0 0 27 0 2 378 3 2 2 4]
[ 0 0 10 0 2 0 434 10 1 156]
[ 4 3 15 1 25 2 9 509 0 3]
[ 0 0 4 0 2 2 2 1 444 6]
[ 1 0 1 1 5 17 42 2 3 309]]
checking accuracy...
(100000, 0.93979999999999997, 0.89480000000000004)
Model saved in file: /tmp/model.ckpt
checking accuracy...
(110000, 0.93879999999999997, 0.8972)
Model saved in file: /tmp/model.ckpt
checking confusion
[[471 0 0 0 0 2 2 0 0 7]
[ 1 560 5 2 0 2 2 14 1 1]
[ 1 0 374 6 2 15 7 4 1 1]
[ 0 0 7 473 0 25 2 3 1 0]
[ 0 2 37 3 511 9 7 24 2 24]
[ 0 0 20 0 1 309 1 4 0 1]
[ 0 0 7 0 0 2 378 2 1 73]
[ 3 1 21 1 17 9 19 498 0 3]
[ 0 0 14 8 1 6 5 0 453 7]
[ 3 0 3 0 3 55 78 1 3 378]]
checking accuracy...
(120000, 0.92920000000000003, 0.88100000000000001)
Model saved in file: /tmp/model.ckpt
checking accuracy...
(130000, 0.94699999999999995, 0.8992)
Model saved in file: /tmp/model.ckpt
checking confusion
[[460 1 0 0 0 0 0 0 0 2]
[ 0 552 2 0 1 0 0 5 0 0]
[ 1 1 386 4 10 7 4 9 2 4]
[ 0 0 13 482 2 19 5 1 4 3]
[ 0 4 17 0 472 1 5 10 0 2]
[ 1 0 30 2 8 381 3 3 2 18]
[ 5 0 14 0 1 5 427 22 2 125]
[ 4 4 17 1 32 2 6 498 0 2]
[ 0 1 7 4 4 2 2 0 449 6]
[ 8 0 2 0 5 17 49 2 3 333]]
checking accuracy...
(140000, 0.93679999999999997, 0.88800000000000001)
Model saved in file: /tmp/model.ckpt
checking accuracy...
(150000, 0.93899999999999995, 0.88880000000000003)
Model saved in file: /tmp/model.ckpt
checking confusion
[[471 0 0 0 0 0 1 0 0 4]
[ 0 553 3 0 0 1 1 2 0 0]
[ 1 2 394 5 17 24 9 15 1 11]
[ 0 0 8 478 1 8 0 0 8 0]
[ 0 5 14 0 492 1 4 20 2 11]
[ 1 0 29 3 3 372 3 8 0 8]
[ 0 0 19 1 2 4 434 21 2 132]
[ 2 3 13 3 14 1 4 482 0 2]
[ 0 0 3 3 0 2 2 0 445 0]
[ 4 0 5 0 6 21 43 2 4 327]]
checking accuracy...
(160000, 0.93979999999999997, 0.88959999999999995)
Model saved in file: /tmp/model.ckpt
checking accuracy...
(170000, 0.93959999999999999, 0.89059999999999995)
Model saved in file: /tmp/model.ckpt
checking confusion
[[470 1 0 1 0 0 1 0 1 3]
[ 0 546 2 0 0 0 0 2 0 0]
[ 1 1 375 6 10 8 4 10 1 5]
[ 0 0 6 476 1 10 4 0 5 1]
[ 0 3 16 1 483 1 2 6 0 4]
[ 0 0 35 2 4 377 4 6 2 5]
[ 0 0 17 0 1 3 419 7 1 103]
[ 3 12 19 3 21 4 9 517 0 3]
[ 0 0 13 4 5 2 2 0 448 5]
[ 5 0 5 0 10 29 56 2 4 366]]
checking accuracy...
(180000, 0.94340000000000002, 0.89539999999999997)
Model saved in file: /tmp/model.ckpt
checking accuracy...
(190000, 0.94279999999999997, 0.90080000000000005)
Model saved in file: /tmp/model.ckpt
checking confusion
[[470 0 0 0 0 1 1 0 0 7]
[ 0 553 3 0 0 0 1 1 0 0]
[ 1 0 390 6 9 14 4 11 1 7]
[ 0 0 3 477 1 6 1 1 4 1]
[ 0 4 15 1 491 1 5 20 4 12]
[ 0 0 31 3 7 399 2 5 0 8]
[ 0 0 19 0 3 0 432 15 3 143]
[ 4 6 18 2 18 1 11 496 0 2]
[ 0 0 7 4 2 1 1 0 446 5]
[ 4 0 2 0 4 11 43 1 4 310]]
checking accuracy...
(200000, 0.94120000000000004, 0.89280000000000004)
Model saved in file: /tmp/model.ckpt
checking accuracy...
(210000, 0.93999999999999995, 0.88119999999999998)
Model saved in file: /tmp/model.ckpt
checking confusion
[[469 0 0 0 0 0 2 0 0 2]
[ 0 554 3 0 1 1 0 5 0 0]
[ 1 0 365 3 10 20 7 14 2 9]
[ 0 0 23 485 4 18 4 7 11 6]
[ 0 3 26 0 480 0 2 17 0 7]
[ 0 1 29 3 6 381 7 6 3 23]
[ 1 0 11 1 2 3 416 6 1 102]
[ 3 5 23 1 20 1 16 494 0 6]
[ 0 0 5 0 6 2 4 0 442 6]
[ 5 0 3 0 6 8 43 1 3 334]]
checking accuracy...
(220000, 0.93500000000000005, 0.88400000000000001)
Model saved in file: /tmp/model.ckpt
checking accuracy...
(230000, 0.94099999999999995, 0.89200000000000002)
Model saved in file: /tmp/model.ckpt
checking confusion
[[474 0 0 1 0 0 3 0 0 8]
[ 0 555 3 0 3 1 1 5 0 0]
[ 1 0 399 4 13 12 20 16 1 8]
[ 0 0 8 479 2 12 5 2 5 1]
[ 0 5 14 2 500 1 3 19 2 8]
[ 0 0 34 1 3 393 6 5 2 10]
[ 0 0 4 0 0 1 389 5 1 115]
[ 2 3 12 2 7 1 6 494 0 4]
[ 0 0 8 4 2 1 6 1 451 7]
[ 2 0 6 0 5 12 62 3 0 334]]
checking accuracy...
(240000, 0.93720000000000003, 0.89359999999999995)
Model saved in file: /tmp/model.ckpt
checking accuracy...
(250000, 0.93720000000000003, 0.88280000000000003)
Model saved in file: /tmp/model.ckpt
checking confusion
[[470 0 0 0 0 0 1 0 0 5]
[ 0 554 2 0 2 0 0 6 0 0]
[ 1 1 384 8 11 19 6 8 2 1]
[ 0 0 4 471 1 9 1 0 3 0]
[ 0 3 13 0 485 0 4 18 1 9]
[ 0 0 30 1 2 367 2 5 1 2]
[ 0 0 11 2 3 6 428 12 2 144]
[ 3 4 29 6 20 4 11 498 0 1]
[ 0 1 8 5 4 2 2 1 449 5]
[ 5 0 7 0 7 27 46 2 4 328]]
checking accuracy...
(260000, 0.94279999999999997, 0.88680000000000003)
Model saved in file: /tmp/model.ckpt
checking accuracy...
(270000, 0.93740000000000001, 0.89600000000000002)
Model saved in file: /tmp/model.ckpt
checking confusion
[[471 0 0 0 0 0 1 0 0 5]
[ 0 556 4 0 1 1 1 9 0 0]
[ 0 0 371 5 9 11 4 4 2 6]
[ 0 0 5 473 1 9 2 0 3 1]
[ 0 2 16 1 463 1 4 19 2 4]
[ 0 1 35 6 7 383 8 5 1 11]
[ 1 0 18 1 2 1 416 8 2 139]
[ 1 4 21 2 37 2 14 501 0 2]
[ 0 0 9 5 3 2 2 0 448 4]
[ 6 0 9 0 12 24 49 4 4 323]]
checking accuracy...
(280000, 0.93540000000000001, 0.88100000000000001)
Model saved in file: /tmp/model.ckpt
checking accuracy...
(290000, 0.93379999999999996, 0.88460000000000005)
Model saved in file: /tmp/model.ckpt
About
Binding parameters to make CNN rotationally invariant
Resources
License
Stars
Watchers
Forks
Releases
No releases published
Packages 0
No packages published