GAN for Audio Samples + MDPs Tutorial + CERN Competition Guide
View this email in your browser


Reproducing research papers is a rewarding way to boost your machine learning skills. But the practice can be deceptively complicated.  If you’re thinking about giving it a try, Matthew Rahtz shares some hard-won insights from his first go at reproducing a Deep Reinforcement Learning paper.

Excited about new ML-based tools for musicians? Check out this WaveGAN code by Chris Donahue, Julian McAuley, and Miller Puckette. It’s a GAN approach designed for use on audio samples; try their groovy in-browser drum sampler demo.

Read on for more of the month's data science notes, highlights and competitions.



Mother's Day Interview: From SWE to Competitive Kaggler in 1 Year

Nicole Finnie and Wendy Kan
Kaggle data scientist Wendy Kan interviews Nicole Finnie, an impressive competitor in the Data Science Bowl who took full advantage of her maternity leave and began learning ML skills a year ago. Read the interview »

Profiling Top Kagglers: Bestfitting, Now #1 in the World

Bestfitting (Shubin Dai)
The new #1 on our leaderboard is a mountain biker from China who joined Kaggle just two years ago. Read Shubin's story here »
Read more recent blog posts


Using MDPs to Optimize Your 2048 Game

Want to learn more about Markove Decision Processes? Take a look at John Lees-Miller's exceptional post on the math behind strategies for the puzzle game 2048.

mltest: New ML Testing Library

Building off the popularity of his Medium post on how to unit test ML code, Chase Roberts recently released a beta ML test library to make the process easier for everyone.

Datasets for Data Cleaning Practice

From Brazilian weather to Halloween candy rankings, Kaggle data scientist Rachel Tatman has gathered a dozen practice datasets that need your TLC. For background, check out Rachel's 5-Day Challenges on Data Cleaning.
Meetup Roundup


Help high energy physics discover and characterize new particles

In this high-powered competition sponsored by CERN, you’re challenged to build an algorithm that reconstructs particle tracks quickly from 3D points left in the silicon detectors. Learn more with this beginner's guide to the competition. Then compete for $25,000 in prizes.

Predict demand for an online classified ad

$35,000 in prizes are on the line in Avito's 4th competition. The challenge: predict demand for an online ad based on its full description and on context cues like geography and historical demand for similar ads. Jump in »
See all active competitions


Data Science for Good Event #2:

Can you help boost repeat donations? That's our 2nd DSFG challenge. These events bring our community together to work on tough social good problems, with datasets that don’t fit the tight constraints of our traditional supervised ML competitions. Learn More »


Job seekers start here

Follow our jobs board and be the first to spot new opportunities. Or create a job listing to reach talent among the nearly 1M data scientists and ML engineers in our global community. Recent job postings include...

Productsup - Data Scientist Developer (Berlin)
Axians - Young Professional (Zaltbommel)
RedShift Zero - Data Scientist (London)
MTU Friedrichshafen GmbH - Solution Architect (Friedrichshafen)
Rijk Zwaan - Onderzoeker Machine Learning (De Lier)
EUMETSAT - Data Engineer (Darmstadt)
... and many more!
Copyright © 2018 Kaggle, All rights reserved.