The joy of feedback or interesting ML projects out there!

One of my greatest pleasures is keeping in touch with our user community and finding out what interesting projects are out there. Here are some small examples from the LAST WEEK. I hope this will encourage further people to contact me!!

Phillip Trotter, ptav: 

This is just a short note to say huge thanks for the GraphChi posts on your blog.

I have  small start up developing complex systems simulation and visualization tools and we are in the early stages of using GraphChi for both feature identification in network graphs and for data validation following simulations. We have potentially a lot of other uses for GraphChi and I am really enjoying getting to know the code base. A big part of that has been your articles and i just wanted to say thank you for the posts - they have been incredibly helpful.

Ben Mabey, Red brain labs

Thanks for all your work on graphlab.  I think it is one of the most interesting ML projects in a long time.  Between graphlab's CF library and libfm I haven't been this excited about recsys since the FunkSVD. :) At my work we're excited about the performance of graphlab since it allows for faster experimentation and cheaper builds.

As a side hobby project I'm trying to develop a recsys for github repositories. The majority of the user feedback is implicit (e.g. commits to repos, opened issues, comments, etc) with the only explicit feedback (staring of a repo) is binary.  Since GraphLab's CF toolkit already implements some of the papers that I had read dealing with implicit feedback it seemed like a great way to get started.

Norman Graham:

thanks for the good work on the GraphLab project. --Norm

Paul Lefkopoulos, 55.com:
Thanks for the work. We are a fresh French startup specialized in web analytics and digital media optimization. One of our main aim is to increase the website conversion rate of our retailing customers. We currently use your tool to build a recommender system for a small retailer in order to increase its cross-sell.

Marilyn Wulfekuhler

 I really appreciate such a responsive and helpful forum!  Thank you, thank you.

Shingo Takamatsu, Sony Japan

I have read your papers on GraphLab (and GraphChi) and implemented a few algorithms on GraphLab (Danny’s blog was great help!).

Xiaobing Liu, Tencent China

I am a senior engineer at Tencent.com.inc which is largest internet companies in
China. I am interested in distributed/parallel large scale machine learning algorithms and applications in search advertising targeting and the relevance of the search engine. So for, I implement parallel logistic regression in batch learning way and in online learning way, which are applied in search ad-ctr prediction and also I implemented some topic models, such as LDA and PLSA all of algorithm bases on MPI. I am quite excited about parallel framework, such as pregel and graphlab.Currently, I am implementing the PLSA and logistic regression base on graphlab.

Zygmunt Zając
I've been using Graphlab mainly for data sport at Kaggle, currently I'm in TOP 100 among 55 thousand registered users. I also write about Kaggle competitions and machine learning in general at fastml.com. I generally like trying new machine learning software and I found Graphlab to be fast, relatively easy to use and comprehensive (a lot of options and algorithms, built-in validation etc.).  
Ayman M Shalaby, University of Waterloo, Canada
I've read heard about the Graphlab project and the project sounds very promising. 
I'm particularly interested in using Graphlab to implement large scale Dynamic Bayesian Networks. 

Steffen Rendle, Author of the libFM collaborative filtering package

I am impressed by the size of the Graphlab project and the efforts you make to scale the algorithms to multiple cores/ machines. I think this will get a very important topic with future increasing parallelism of computers.






And how can we forget our mega collaborator JustinYan?

I am Qiang (Justin) Yan ,a master student from the Chinese Academy of Sciences. My main focus is large scale data mining and Collaborative Filtering. I was working as an intern to build Recommender system for Baidu and Hulu in last couple of years. Recently I focus on implementing Collaborative Filtering Models on GraphLab and GraphChi.