epagoge.de - Fabian Sinz

Datajoint

I joined the developer team for datajoint-python. Datajoint is a object relational mapper (ORM) for mysql that makes integration of different experimental data sources, their analysis, and retrieving results really easy. What I really like about datajoint is that it uses relational algebra to keep your analyses consistent. Dependencies are modeled through foreign key constraints. If a particular parameter is changed or some piece of data is deleted, all depending analyses are automatically deleted as well.

Datajoint is available for Python and Matlab, with cross language compatibility. In the Python version, we really though hard on how to structure things such that they are easy to handle, but powerful and flexible at the same time. I feel we didn't do such a bad job and I am pretty excited about using it in the future. I will write short blog entries on datajoint.github.io to demonstrate special use cases such as distributed computing or initializing a table with google spreadsheets.

Datajoint-python is available via pip, so check it out. Or even better, fork it on github and contribute.


Baylor College of Medicine

Since mid of April 2015, I am a member of the lab of Andreas Tolias at the Baylor College of Medicine in Houston, TX. I am looking forward to a lot of exciting research.


6th G-Node Winter Course on Neural Data Analysis

I had the opportunity to teach one course day at the 6th G-Node Winter Course on Neural Data Analysis. Thanks to the students, the other faculty, and all the organizers I had a very enjoyful time. The lecture slides can be downloaded here.


Least Informative Dimensions

Our paper Least Informative Dimensions got accepted at NIPS 2013. Read more about it here.



What is the Limit of Redundancy Reduction with Divisive Normalization?

Our paper about about the Limit of Redundancy Reduction with Divisive Normalization got accepted a Neural Computation. Read more about it here.