3.6. Auto-encoder with Conv2D¶
This notebook demonstrates using a Conv2D network in an autoencoding task with the MNIST dataset.
In [1]:
import conx as cx
Using Theano backend.
Conx, version 3.6.0
First, we make a network. We will work with a 3 dimensional input from MNIST, but a flat target vector.
In [2]:
net = cx.Network("Auto-Encoding with Conv")
net.add(cx.Layer("input", (28,28,1)),
cx.Conv2DLayer("Conv2D-1", 16, (5,5), colormap="gray", activation="relu"),
cx.MaxPool2DLayer("maxpool1", (2,2)),
cx.Conv2DLayer("Conv2D-2", 132, (5,5), activation="relu"),
cx.MaxPool2DLayer("maxpool2", (2,2)),
cx.FlattenLayer("flatten"))
net.add(cx.Layer("output", 28 * 28, vshape=(28,28), activation='sigmoid'))
net.connect()
In [3]:
net.compile(error="mse", optimizer="adam")
We load the MNIST dataset and examine the shapes of the inputs and targets.
In [4]:
net.dataset.get("mnist")
In [5]:
net.dataset.info()
Dataset: MNIST
Original source: http://yann.lecun.com/exdb/mnist/
The MNIST dataset contains 70,000 images of handwritten digits (zero to nine) that have been size-normalized and centered in a square grid of pixels. Each image is a 28 × 28 × 1 array of floating-point numbers representing grayscale intensities ranging from 0 (black) to 1 (white). The target data consists of one-hot binary vectors of size 10, corresponding to the digit classification categories zero through nine. Some example MNIST images are shown below:
MNIST Images
Information: * name : MNIST * length : 70000
Input Summary: * shape : (28, 28, 1) * range : (0.0, 1.0)
Target Summary: * shape : (10,) * range : (0.0, 1.0)
Because this is an auto-encoding task, we wish that the targets were the same as the inputs:
In [6]:
net.dataset.set_targets_from_inputs()
WARNING: network 'Auto-Encoding with Conv' target bank #0 has a multi-dimensional shape, which is not allowed
However, that gives a warning. Conx does not allow targets to have a shape, so we flatten them:
In [7]:
net.dataset.targets.reshape(28 * 28)
In [8]:
net.dataset.targets.shape
Out[8]:
[(784,)]
In [10]:
net.dashboard()
Just to test our design, we chop the majority of patterns, leaving only 100.
In [11]:
net.dataset.chop(69900)
And save 10% for testing/validation:
In [12]:
net.dataset.split(0.1)
In [13]:
net.reset()
net.train(50)
========================================================
| Training | Training | Validate | Validate
Epochs | Error | Accuracy | Error | Accuracy
------ | --------- | --------- | --------- | ---------
# 50 | 0.02086 | 0.00000 | 0.02440 | 0.00000