Blog with Keras news, tutorials, and demos.
Read at: blog.keras.io
Generated with Pelican, hosted on Github pages.
To contribute an article: send a Pull Request to the content branch. We welcome anything that would be of interest to Keras users, from a simple tutorial about a specific use case, to an advanced application demo, and everything in between.
While reproducing the examples from the blog for a lecture I realised that the losses dont add up.
Plain 784->32->784 indeed gives 0.10 easily and can even lower.
Second step is the activity regularization.
from keras import regularizers
encoding_dim = 32
input_img = Input(shape=(784,))
# add a Dense layer with a L1 activity regularizer
encoded = Dense(encoding_dim, activation='relu',
activity_regularizer=regularizers.activity_l1(10e-5))(input_img)
decoded = Dense(784, activation='sigmoid')(encoded)
autoencoder = Model(input=input_img, output=decoded)
# this model maps an input to its encoded representation
encoder = Model(input=input_img, output=encoded)
# create a placeholder for an encoded (32-dimensional) input
encoded_input = Input(shape=(encoding_dim,))
# retrieve the last layer of the autoencoder model
decoder_layer = autoencoder.layers[-1]
# create the decoder model
decoder = Model(input=encoded_input, output=decoder_layer(encoded_input))
autoencoder.compile(optimizer='adadelta', loss='binary_crossentropy')
Going by the book I cannot get anything better than: 0.26 after 300+ epochs
Has keras underling code changed since the post or something?
Im using theano with latest keras.
Currently the feed at https://blog.keras.io/feeds/all.rss.xml has links like this:
<link>
//blog.keras.io/user-experience-design-for-apis.html
</link>
However some platforms (like Slack, and various email clients, and some versions of some mobile browsers) do not handle these links as links.
Therefore, given that the site is on HTTPS, it would be more robust to simply use the full URL with protocol (https:) included.
<link>
https://blog.keras.io/user-experience-design-for-apis.html
</link>
See also: getpelican/pelican#1532
Hi,
I want to create a new network out of two existing networks (models) A and B using Keras with tensorflow. Architecture of new network is like this
Input -> A -> B -> Output
The output of A has shape of (15, 500) and input of B has a shape of (1000). I have a conversion method that takes the input of shape (15, 100) and call predict of a completely different network (C), and return the output with the shape of (1000).
I am assuming that I have to introduce a Lambda layer which will use my conversion method to convert the output of A to a format required for B. But, when I tried that, I get error
"TypeError: unsupported operand type(s) for /: 'Dimension' and 'float'_" on calling predict in the conversion method.
This is the conversion method
.def convert(x):
.....C = load_model("path/to/the/network/C.h5")
.....return C.predict(x)
I am not sure if this is the right way of doing, if it is, then why is this error. If it is not the right way, then, what should be my approach.
Keras version: 2.0.1
Tensorflow version: 1.0.1
Thanks,
The Atom feed under the <head> tag on http://blog.keras.io/ points back to http://blog.keras.io/ instead of an actual Atom feed.
If you add FEED = FEED_ATOM = 'feed.xml' to your Pelican config, it should automatically generate the feed.
In the article https://blog.keras.io/keras-as-a-simplified-interface-to-tensorflow-tutorial.html, I found I had to add
init_op = tf.global_variables_initializer()
sess.run(init_op)
before calling train_step.run on the MNIST data.
(I don't see a source file for this article in the content folder, so I can't send a pull request.)
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
Django
The Web framework for perfectionists with deadlines.
Laravel
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
Microsoft
Open source projects and samples from Microsoft.
Google
Google ❤️ Open Source for everyone.
Alibaba
Alibaba Open Source for everyone
D3
Data-Driven Documents codes.
Tencent
China tencent open source team.