initial commit

Author: harsh306

README.md

@@ -0,0 +1,20 @@
+# Datasets
+
+A curated list of Datasets that can be used to implement/test curriculum learning. 
+
+| Datasets        | Paper/Project link  |
+| ------------- |:-------------:| 
+|    GQN by DeepMind  | [Link](https://github.com/deepmind/gqn-datasets) | 
+| Deep Pentomino      | [Link](https://github.com/caglar/Arcade-Universe) |
+| Pentomino by DeepMind | [Link](https://github.com/deepmind/deepmind-research/tree/master/PrediNet)|
+| C-SMOTE Generator| [Link](c_smote_generator/README.md) - [ Link2](https://digitalcommons.wpi.edu/etd-theses/1256/)|
+|Parity task | [Link](parity_task/README.md) - [Link2](https://openreview.net/forum?id=r1G4z8cge)|
+| Progressive GANs Generator | [Link](https://github.com/tkarras/progressive_growing_of_gans#preparing-datasets-for-training)|
+| Shapes 2D  Generator | [Link](https://github.com/matterport/Mask_RCNN/blob/master/samples/shapes/shapes.py) |
+| Shapes Four Kaggle | [Link](https://www.kaggle.com/smeschke/four-shapes)|
+| Shapes  3D| [Link](https://github.com/deepmind/3d-shapes)|
+| Shapes Flying | [Link](https://github.com/ferreirafabio/FlyingShapesDataset)|
+| Scikit-learn data Generator|[Link](https://scikit-learn.org/stable/modules/classes.html#samples-generator)|
+|Local-minima-propositions | [Link](https://openreview.net/forum?id=Syoiqwcxx)|  
+|MNIST |[Link](http://yann.lecun.com/exdb/mnist/)|
+|CIFAR |[Link](https://www.cs.toronto.edu/~kriz/cifar.html)|

c_smote_generator/README.md

@@ -0,0 +1,12 @@
+C-SMOTE is an algorithm that generates data synthetically given the modes 
+and support of original datasets. For more information we suggest reading this Thesis.
+
+[Parameter Continuation with Secant Approximation for Deep Neural Networks](https://digitalcommons.wpi.edu/etd-theses/1256/)
+
+## Grid Dataset 
+![grid](grid_smote%20(1).gif)
+
+## Sine-Wave Dataset
+![sine](sine_smote%20(4).gif)
+
+## Script

c_smote_generator/grid_smote (1).gif

@@ -0,0 +1,22 @@
+Parity task Generator
+
+What is parity task?\
+A parity bit, or check bit, is a bit added to a string of binary code to ensure that the total number of 1-bits in the string is even or odd [Wikipedia](https://en.wikipedia.org/wiki/Parity_bit)
+
+40-bit parity task dataset used in [Mollifying Networks](https://openreview.net/forum?id=r1G4z8cge)
+
+```
+def parity_batch(input_length, batch_size):
+    xs = [np.random.randint(0, 2, input_length) for _ in range(batch_size)]
+    xs.append(np.ones(shape=input_length, dtype=int))
+    ys = [[0] if np.sum(x) % 2 == 0 else [1] for x in xs]
+    return xs, ys
+```
+
+```
+import numpy as np
+np.random.seed(seed=config.seed)
+d_dim = config.parity_length
+X, y = parity_batch(config.parity_length, 2000)
+X_val, y_val = parity_batch(config.parity_length, 500)
+```