I’m a bit fascinated about KI lately. What finally made me interested was a report about George Hotz using machine learning to build a self driving car. And a deeply respected Friend of mine, Flavio Trolese, working on an “Cloud Power for Image KI” startup.

The ide is that working as a coder, I really appreciate how source code is managed by git nowadays. It’s a bit of a learning curve, but then it gets intuitive and very productive in managing simultanously changing code.

One of the core concepts of git is to “merge” the codes for two features into a new code base that contains both features. When changes not interfere they can be merged automatically. But when both changes touch the same lines of code, manual interventions is needed.

Most of these interventions are booring. The way the “conflicts” are presented is not intuitive for most of mankind. There’s tools, but anyway, solving such issues take developer time.

I think we should be able to build tools, using state-of-the-art open source technologies, automating most of the conflicts.

I try to blog about my progess into this. Maybe I succeed:-)

Some links for today:


A framework I’m currently giving a try. It includes some nice tutorials, like MNIST, the “hello world” of computer vision.

Awesome Deep Learning

One of many awesome lists. Lost of links to tools, tutorials, special areas…

Gmail Blog: Computer Respond to this Email

Similar use case: learn from millions of (merges|mail threads), and figure out an appropriate (resolution|reply).

Building Program Vector Representations for Deep Learning

A Chinese Paper I’ll give a try reading:-)


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s