Winning solution for Yelp! Business Prediction Contest

Just heard from my mega collaborator Justin Yan, now working in Alibaba, that his lab mates won the 1st place in the Yelp! business prediction contest. The team is composed of students from the Chinese Academy of Science, headed by Yuyu Zhang.

Brief solution method is explained here (see middle post from Brick Mover).

I will try to extract some additional tips from the winning team and report them here soon.

MLConf Machine Learning Workshop- Nov 2013 in SF

Our event organizers Courtney & Shon Burton are organizing a machine learning workshop mid November (Nov. 15) in SF.

Preliminary agenda is here. Preliminary speakers are Xavier Amatrain (Netflix), Ted Willke (Intel Labs), Jake Mannix (Twitter), Joey Gonzalez (GraphLab), Eric Bieschke (Pandora), Ameet Talwalkar (Berkeley) and others.

All of my blog readers are welcome to use this discount code.

Big Learning 2013 Workshop

My collaborator Joey Gonzalez is organizing this year (again) the NIPS Big Learning 2013 workshop.
The workshop will take place early December 2013 in Lake Tahoe. Submission deadline is October 9th 2013.

Some of this year's topics:

• Distributed algorithms for online and batch learning
• Parallel (multicore) algorithms for online and batch learning
• Theoretical analysis of distributed and parallel learning algorithms
• Implementation studies of large-scale distributed inference and learning algorithms --- challenges faced and lessons learnt
• Database systems for Big Learning --- models and algorithms implemented, properties (availability, consistency, scalability, etc.), strengths and limitations

MLBase + Spark talk

I got his video talk from Senthil Gandhi, a senior data scientist at Firat Retail Labs. It is a video of MLBase project on top of Spark


The talk is by Ameet Talwalker and Evan Sparks from the Berekely AMP Lab. Video was released a couple of weeks ago, of a talk given at Twitter on Aug 6, 2013. Very recommended!

On a related note, I just got from my collaborator Joey Gonzalez, our man at the AMP lab, a link to the recent AMP camp. The talks videos will be released soon.

A guide for python frameworks for Hadoop

I got this interesting link from J D Chen, my colleague at Renren (Chinese Facebook).
It is a talk by Uri Laserson from Cloudera who is comparing different python high level programming on top of Hadoop.   Lecture slides are available here. And here is a detailed blog post.

As we demoed in GraphLab workshop, we are also working on a python interface to GraphLab. Stay tuned!

Open PhD/Postdoc positions for large scale graph analytics

My colleague Oded Margalit from IBM Haifa sent me a note about an open position for a joint IBM/Technion funded PhD or Postdoc position in large scale distributed systems research.
Anyone who is interested is welcome to apply (the position is open only for researchers who where outside of Israel in the last 3 years).

Another open PhD position in KTH (Stockholm) for social network and graph processing researcher, I got from my previous collaborator Sarunas Girdzijauskas.

Amazon's CTO writes about GraphLab in his blog

I got the following from a former colleage of mine: Barak Merimovich from Gigaspaces. It seems that Amazon's CTO, Werner Vogel have written recently about GraphLab in his blog:

"I want to kick off the fall readings not with an historical paper but with two that detail GraphLab, an excellent framework for high performance machine learning that originally has been built by the Carlos Guestrin. GraphLab has been used to build several different data mining and graph processing toolkits and applications. The research in the papers has been performed on Amazon EC2. Instructions for running your own GraphLab Cluster on EC2 can be found here ..."

Additionally, I learned today from our man in O'Reilly Media, Ben Lorica, that has mentioned GraphLab in his blog post.