🌊 Water you doing, my friend?

it’s finally cooling down over here. The sun is nice and all, but above 20°C, my brain just stops working. But that shouldn’t stop us from finding some awesome machine learning and some Python this week! Let’s dive right in!

The Latest Fashion

• The creator of XKCD started playing around with DALL-E 2 exploring Pokemon cards throughout the ages.
• Matplotlib has a Mosaic API that seems really cool for subplots I didn’t know about!
• Truss promises to serve ML models without boilerplate. Enticing!

Got this from a friend? Subscribe here!

My Current Obsession

The heat! Luckily today it cooled down, but the heat has been brutal! On the plus side, I found an app that actually uses the ECMWF data. And with “found”, I mean that some people in ECMWF use it, because we’re proud of our forecasts. It’s called windy.com and looks very pretty. (Not an official product from us, they just ingest our data.)

I also started going bouldering again. I was always afraid, since I have gained so much weight and I messed up my elbows climbing before. But all that strength training and going slow seems to make a difference! Second time and it seems alright. Very happy to be back on the wall!

Euroscipy published their schedule now and my tutorial “Increase citations, ease review & collaboration – Making machine learning in research reproducible” will be on the 29th at 10:30. Nice and early, for me to enjoy the conference after without worrying. They couldn’t have known, but that’s just lovely for me. Even after so many presentations, talks, and videos, I still get nervous every time, so having that out of the way is fantastic. I’m really looking forward to it and hope I can reach my ambitious goal during the session. Let me know if you’re coming to beautiful Switzerland! It’s one of my favourite conferences.

Thing I Like

It’s hot, I have two things on my mind. Ice cream and my tower fan. It’s not a fancy Dyson, although one can dream, but it moves the air and I’m thankful for it.

Hot off the Press

I wrote a long blog post about changing careers from Oil & Gas to have a positive impact on the world instead. I was nervous about publishing it, as it’s quite personal and dear to my heart, but I think it was very well received.

Earlier I wrote a post about books that are great in natural language processing (NLP).

Machine Learning Insights

Last week I asked, “What is your favourite machine learning algorithm and can you explain it in under 1 minute?”, and here’s the gist of it:

I love a good neural network. It’s where I got my start in machine learning and it’s one of the main tools I use.

With dense neural networks I like to go from the basic linear regression equation y = m • x + b from school. We have the input data x and target data y, with m being the weights we learn and b being a bias that is also learnable. In dense neural networks, we connect all the inputs in x to each neuron, so we’re dealing with vectors instead that sum all incoming weighted data at each Neuron y = Σ m • x + b. Since neural networks shouldn’t just be linear regression models, we need something that makes this equation non-linear. Normally, we simply define an activation function σ so that y = σ( Σ m • x + b). Now we stack a bunch of neurons to build up a layer, so we get additional processing power. Finally, we get to connect layers to each other, by feeding the output of one layer into the other. So our x in layer 2 is the y in layer 1, which gets to look really cool y = σ( Σ m' • σ( Σ m • x + b) + b'). During training, we can easily pass data through our system of linear algebra and when we calculate the error, we can use the chain rule to correct the system. It’s beautiful in its simplicity, really.

It’s also why people call deep learning the most lucrative application of the chain rule ever.

Data Stories

Time is relative.

Time travelling even more so. How far do you get in 5 hours by train starting from different locations in Europe?

I found this map fascinating, and it’s interactive! Clicking around and finding the different hubs for different countries is a bit of fun for a slow afternoon!

Source: Chronotrains

Question of the Week

• What is boosting and can you name examples of its usage?

Post them on Twitter and Tag me. I’d love to see what you come up with. Then I can include them in the next issue!

Tidbits from the Web

• So… what’s actually going on with Google and Facebook’s hiring freezes?
• This story of Casper suing a mattress reviewing site over preferring a competitor is eye-opening and somewhat not surprising. The end, however, is.
• Origami for blind people? A whole book of it even? Fascinating and I love it!

A secret one for the people reading to the end. I was thinking about ramping up the referral rewards. Maybe mugs, or a shirt, maybe some digital goods. What is something you think would make your day getting for showing my newsletter to friends and colleagues? Simply reply to the email. It’ll land in my inbox! I’d really appreciate the help!