TensorFlow, Deep Learning, and Modern Convolutional Neural Nets, Without a PhD (Cloud Next '18)

The hottest topics in computer science today are machine learning and deep neural networks. Many problems deemed ""impossible"" only five years ago have now been solved by deep learning: playing GO, recognizing what is in an image, or translating languages. Software engineers are eager to adopt these new technologies as soon as they come out of research labs, and the goal of this session is to equip you to do so. This session will focus on the newest developments in image recognition and convolutional neural network architectures and give you tips, engineering best practices, and pointers to apply these techniques in your projects. No PhD required.

Speaker: Martin Görner

re_ty: Publish; product: TensorFlow - General; fullname: Martin Görner; event: Google Cloud Next 2018;

[thomasfel]

As always, great talk... Please just continue doing this Mr GÖRNER 👌
btw the cartoon are great 😉

[victorariasvanegas]

In minute 16:28 there are several mistakes in my opinion.
YY = tf.reshape(Y3, shape=[-1, 5 * 5 * 32])
...
Y1 = tf.layers.dense(Y4, 2, activation=None)

If not, I apologize.

[huqiao]

I was able to train the model and to deploy it locally using Docker, instead of using the ML Engine. But I am having difficulties to connect the webui to direct the map images to my local TensorFlow Serving instance. I can see that the function analyze() in dash.js calls mlengine.projects.predict with name:model_url and resource:body(images). How can I modify this part of code to call my local RESTful API?