Skip to content

DSB #103


Friday is finally here and also the DSB! For me the most interesting link in this DSB is the one about hardware lottery in Computer Science. Also for those who are into Rasa, the link in Education gives you opportunity to take a course on Udemy for free. But please, do not use it if you are not truly interested. Thanks 😉

And as always, enjoy your reading.

Analytical – How to recognize a good estimator? Learn about unbiasedness, consistency, asymptotic normality, efficiency and robustness. – Use Mindee’s API to parse and extract data from receipts. Developer account is for free. (rcmd by reader) – Do you like tidyverse, but you want be as fast as data.table? Then tidytable is exactly what you need! (rcmd by reader)

Computer Science & Science – Impressive long reading about hardware lottery, a problem when your solution is the best one there is just because it suits the hardware and software you are using. It is written by Sara Hooker, a researcher at Google Brain. – A brief overview of AWS services for those who do not have any idea yet. (rcmd by reader) – We all know that GPU is better for deeplearning than CPU, but why?

Graphs and Visualizations – ggplot is the best – convince me otherwise – now you can even export editable plot to PowerPoint. (rcmd by reader) – Pěkný dashboard od, který Vám ukazuje současné kapacity nemocnic. Na konci stránky najdete odkaz na data. – Geographical plots with plotly and cufflinks in Python.

Business and Career – Microsoft is licensing GPT-3 by OpenAI (more about it also here) and the question is what does it mean for OpenAI itself or for other researchers? (rcmd by reader) – Older article about innovative cultures, how to create one and which principles are essential. (rcmd by reader) – Naprosto bizarní zpráva o zesnulé ženě, která do bank skrze směnárny dostala 12 mld. Kč…

Pop – Last year in DSB #75 we shared a report about AI in 2019. Now you can read updated version of this comprehensive report for current funny year 2020. (rcmd by reader) – AI predictions who will be the next James Bond. (rcmd by reader) – Rozhovor s profesorem z ČVUT Vladimírem Maříkem, expertem na robotiku, který zde hovoří o AI, robotizaci a návazných tématech nejen pohledem ČR. (rcmd by reader)

Education –  Please, only for those whore are really interested in Rasa, you can use this free ticket to take a course on Udemy. It is valid only for two days, so don’t miss this opportunity, if you are interested! First come, first served. (rcmd by reader) – Make your own computer science bachelor degree with these free courses, I bet you recognize many of them. Author of the list explains the idea here in the video. (rcmd by reader) – Intro to recommenders in TensorFlow.

Data & Data Mining – Fast data transformation in R with collapse, C/C++ based package. (rcmd by reader) – Data is the new oil, whatever, but what is truly important in case of data in your company? – Just introducing coax, a modular reinforcement learning Python package.

Papers & Books – Essential papers from RecSys 2020, conference on recommender systems. – Authors review the different applications for mobile phone data in guiding and evaluating COVID-19 response.

Behind the Fence – Data Science Manager in New York, USA.

Joke – I press the Shift, though…

Hacking Time

It has been a long time since we had the hacking time! This task was recommended by reader.  


  • just two lists:

a = [“__asd”, “sdfsd”, “__Asfda”, “__asfsdfsdf”, “dfs”, “dfd”]

b = [“O”, “X”, “O”, “X”, “O”, “O”]

Problem description:

  • It is a “conditional reduce”, when the dictionary elements “a” and “b” are merged to words/letters.
  • The merge takes place between the element i and ii, unless ii begins with “__”.


[(‘asdsdfsd’, [“O”, “X”]), (‘Asfda’, [“O”]), (‘asfsdfsdfdfsdfd’, [“X”, “O”, “O”])]

Be First to Comment

Leave a Reply