Skip to content
Snippets Groups Projects
Commit 02431f70 authored by Julien Dejasmin's avatar Julien Dejasmin
Browse files

add visualizations results

parent 61d7e641
Branches
No related tags found
No related merge requests found
Showing
with 32 additions and 25 deletions
No preview for this file type
......@@ -31,7 +31,7 @@ The MNIST database of handwritten digits, available from this [link](http://yann
## Results on MNIST:
### Loss/ACC:
### Loss/ACC: with 10 epochs.
| Models: 2 conv layers (29k parameters) | Loss | Accuracy (%) |
|:-----------------------------------------------------------------: |:--------------: |:--------------: |
| No binary models | **0.0341** | **98.79** |
......@@ -73,20 +73,25 @@ The Omniglot data set is designed for developing more human-like learning algori
[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1Sd1zvImf4UzTiix7mnI6Hzk7vkeKNSLB#scrollTo=XVgZBwOCIdl0)
## Results on MNIST:
### ACC:
## Results on Omniglot:
### ACC: with this repository with 10 epochs.
| Models: matching Network (MN) [4] | Accuracy (%) | Accuracy (%) |
|:-----------------------------------------------------------------: |:--------------: |:--------------: |
| k-way | 5 | ** |
| n-shot | 1 | ** |
|:-----------------------------------------------------------------: |:--------------: |:--------------: |
| No binary MN | 1 | ** |
|:-----------------------------------------------------------------: |:--------------: |:--------------: |
| binary MN: first conv | ? | ? |
| binary MN: second conv | ? | ? |
| binary MN: third conv | ? | ? |
| binary MN: fourth conv | ? | ? |
| Models: matching Network (MN) [4] | Accuracy (%) |
|:-----------------------------------------------------------------: |:--------------: |
| k-way | 5 |
| n-shot | 1 |
|:-----------------------------------------------------------------: |:--------------: |
| No binary MN | **84.4** |
|:-----------------------------------------------------------------: |:--------------: |
| binary MN: first conv | **79.6** |
| binary MN: second conv | **79.6** |
| binary MN: third conv | 64.8 |
| binary MN: fourth conv | 53.6 |
### Heatmap:
heatmap No binary network, conv layer1:
![heatmap binary network conv1|150x150](results/Omniglot_results/heatmapbinary_MN_first_conv_conv1.png)
# References:
......
......@@ -4,10 +4,10 @@ import os
PATH = os.path.dirname(os.path.realpath(__file__))
# local DATA_PATH
DATA_PATH = '/home/julien/PycharmProjects/thesis/work/Pytorch/MNIST_Binary_V2/data/'
# DATA_PATH = '/home/julien/PycharmProjects/thesis/work/Pytorch/MNIST_Binary_V2/data/'
# colab DATA_PATH
# DATA_PATH = 'data/'
DATA_PATH = 'data/'
EPSILON = 1e-8
......
No preview for this file type
No preview for this file type
No preview for this file type
No preview for this file type
No preview for this file type
This diff is collapsed.
epoch,categorical_accuracy,loss,lr,val_1-shot_5-way_acc,val_loss
1,0.778,0.9533519837545685,0.001,0.8,0.616737681295656
2,0.8622666666666668,0.4550229909799085,0.001,0.87,0.43833875964832636
3,0.9007999999999998,0.320360649811288,0.001,0.896,0.3457465486303751
4,0.9274666666666661,0.25274506511145584,0.001,0.894,0.3231195524051118
5,0.9230666666666666,0.24372688643086973,0.001,0.91,0.33259988474469887
6,0.9389333333333331,0.19657207039475327,0.001,0.89,0.2970636524840366
7,0.9374666666666661,0.18138220079964518,0.001,0.898,0.3222561237411739
8,0.9471999999999996,0.1596292485957262,0.001,0.926,0.2700234619845729
1,0.7026666666666667,2.104287913513441,0.001,0.44,1.5924072702723575
2,0.7920000000000001,1.1609025717093833,0.001,0.62,1.202949864179473
3,0.7560000000000001,1.0870123655266073,0.001,0.74,0.769391296713978
4,0.752,0.8491624999140818,0.001,0.82,0.5381571792938985
5,0.768,0.7014528157320702,0.001,0.76,0.8621403412271347
6,0.8013333333333333,0.5696974590719571,0.001,0.84,1.0207692265302541
7,0.8133333333333332,0.5612485399407076,0.001,0.86,0.4157242334214697
8,0.7973333333333333,0.5550898496917309,0.001,0.76,0.7614003585715885
9,0.8213333333333332,0.451805199095923,0.001,0.8,0.6756177308612625
10,0.8573333333333333,0.4525841653977821,0.001,0.84,0.43197904722653646
results/Omniglot_results/plot_acc_loss/No_Binary/acc_model_MN_no_binary.png

19.1 KiB | W: | H:

results/Omniglot_results/plot_acc_loss/No_Binary/acc_model_MN_no_binary.png

17.4 KiB | W: | H:

results/Omniglot_results/plot_acc_loss/No_Binary/acc_model_MN_no_binary.png
results/Omniglot_results/plot_acc_loss/No_Binary/acc_model_MN_no_binary.png
results/Omniglot_results/plot_acc_loss/No_Binary/acc_model_MN_no_binary.png
results/Omniglot_results/plot_acc_loss/No_Binary/acc_model_MN_no_binary.png
  • 2-up
  • Swipe
  • Onion skin
results/Omniglot_results/plot_acc_loss/No_Binary/loss_model_MN_no_binary.png

15 KiB | W: | H:

results/Omniglot_results/plot_acc_loss/No_Binary/loss_model_MN_no_binary.png

18.2 KiB | W: | H:

results/Omniglot_results/plot_acc_loss/No_Binary/loss_model_MN_no_binary.png
results/Omniglot_results/plot_acc_loss/No_Binary/loss_model_MN_no_binary.png
results/Omniglot_results/plot_acc_loss/No_Binary/loss_model_MN_no_binary.png
results/Omniglot_results/plot_acc_loss/No_Binary/loss_model_MN_no_binary.png
  • 2-up
  • Swipe
  • Onion skin
No preview for this file type
No preview for this file type
No preview for this file type
No preview for this file type
No preview for this file type
No preview for this file type
0% Loading or .
You are about to add 0 people to the discussion. Proceed with caution.
Please register or to comment