Google TensorFlow released in the open source!!!

According to Google, the best machine learning suite in the world, TensorFlow, has now been released into the public as an open-source project on GitHub. Well done, Google! “TensorFlow is an open source software library for numerical computation using data flow graphs. Nodes in the graph represent mathematical operations, while the graph edges represent the…

Docker & Rocker

The current talk of the town, seems to be a virtualization method for applications which is called Docker which is (was) based on Linux containers, but can now also be used with virtual machines like VirtualBox to run on Mac OS and Windows OS with minimal performance loss. The docker approach has also been applied…

Organizing multiple tables in R/knitr/LaTeX (Sweave) documents

When reporting results in automated documents using literate programming as implemented in R/knitr/LaTeX, arranging multiple floating objects (figures and tables) can quickly become a problem, because of LaTeX’s typesetting defaults which waste space or throw errors, because LaTeX cannot place the floats without violating the constraints, leading to too many floats in the buffer. Here…

DMML classics

Just to remind readers that the following textbooks on data mining and machine learning are classics and freely available. After downloading them to your ebook reader, they should be readable without soon tiring the eyes … Hastie, T., Tibshirani, R., & Friedman, J. (2013). The Elements of Statistical Learning: Data Mining, Inference, and Prediction, Second…

Chrome | screencastify

Screencastify is a very elegant, easy use software to create screencasts (Desktop + webcam), which is nicely integrated into the CHROME browser as an extension and to Google Drive as storage. ¬†Screencasts are mostly tutorials on computer or software usage, but can be used for presentations in the broadest sense which combine screen and audio.…

AstraZeneca-Sanger Drug Combination Prediction DREAM Challenge

Any volunteers? Who wants to join me? “To accelerate the understanding of drug synergy, AstraZeneca has partnered with the European Bioinformatic Institute, the Sanger Institute, Sage Bionetworks, and the distributed DREAM community to launch the AstraZeneca-Sanger Drug Combination Prediction DREAM Challenge. This Challenge is designed to explore fundamental traits that underlie effective combination treatments and…

Hamiltonian Monte Carlo in easy words

If you are playing around with R/stan and struggling to get your head around the HMC/NUTS algorithm, I highly recommend the following blog by Prof. John Thompson which is a very educational description of the underlying principles (using Stata though): Stan with Stata, Part V: Kicking a marble around in a bucket Stan with Stata,…