1 min read16th Jul 20174 comments

9

After signing up for this post, those of us that want to study machine learning have made a team.

In an effort to actually get high returns on our time we won't delay, and instead actually build the skills. First project: work through Python Machine Learning by Sebastian Raschka, with the mid-term goal of being able to implement the "recognizing handwritten digits" code near the end.

As a matter of short term practicality currently we don't have the hardware for GPU acceleration. This limits the things we can do, but at this stage of learning most of the time spent is on understanding and implementing the basic concepts anyway.

Here is our discord invite link if you're interested in joining in on the fun.

 

September 14th 2017 edit:

Wasn't enough activity to keep it going. The book we chose also focused on scikitlearn rather than teaching how some of the algorithms are programmed after ~chapter 2 so I think that was part of the problem, but the main failing I suspect is that I personally did not push forward hard enough and got distracted by other projects.

So alas, this effort did not succeed. Feel free to check out the discord to find the people interested in it, and maybe look at the logs to see where we went wrong ;)

New Comment
4 comments, sorted by Click to highlight new comments since: Today at 5:55 PM

I am in the group. I am getting started tomorrow!

As a matter of short term practicality currently we don't have the hardware for GPU acceleration. This limits the things we can do, but at this stage of learning most of the time spent is on understanding and implementing the basic concepts anyway.

For what you're doing, GPU stuff probably doesn't make that big of a difference. Convolutional networks will train and run faster, but a digit recognition network should be tiny and fast anyway.

+1 on actually doing this.

this looks like a fun tech to play with:

"We're teaching computer characters to learn to respond to their environment without having to hand-code the required strategies, such as how to maintain balance or plan a path through moving obstacles. Instead, these behaviors can be learned."

DeepLoco: Dynamic Locomotion Skills Using Hierarchical Deep Reinforcement Learning

http://www.cs.ubc.ca/~van/papers/2017-TOG-deepLoco/

https://www.eurekalert.org/pub_releases/2017-07/uobc-bst073117.php