Tuesday, September 29, 2015

Kudus: Cloudera's answer to Apache Parquet

I got this from my colleague Guy Rapaport: Kudu.io a framework for fast querying on top of Hadoop HDFS. And here is a blogpost discussing the project importance.

Saturday, September 26, 2015

O'Reilly Data Science Salary and Tools Survey

I got this from my colleague Assaf Spanier from Correlor: O'Reilly Data Science and Salary Survey.
Salaries look on the low side to what I see but the used tool captures well industry trends.

Wednesday, September 23, 2015

WhetLab bought by Twitter

Slightly late, but I just heard from Alessando Vitale (CEO Optimist AI) about WhetLab. An interesting effort to make deep learning configuration easier. Unfortunately they where immediately bought by Twitter. Here is a video which explains what they did (before being bough by Twitter which of course shutdown this activity):

Thursday, September 10, 2015

Apache Singa - new distributed deep learning framework

I learned today from my friend Assaf Araki from Intel about Apache Singa project which had a workshop at VLDB 2015. The project is maintained by a few Singaporean universities. It is a C++ platform (with Python binding) which already implements round 10 different deep learning algorithms. It has support for different ways of computing distributed SGD, both synchronous and asynchronous like Hogwild! and ParameterServer.