🧐 Monopoly startups are doomed – There's no real money in it

it’s been a while! My vacation made me forget the time, as it should, so I completely missed last week. Apologies! But I’m back at my desk now. Full of energy. Ready to take on the world of machine learning! So let’s dive in.

The Latest Fashion

• You know the advice to not “copy and paste commands” because it’s unsafe and you still always do? Shellfirm tries to secure your shell for risky command patterns like deleting everything with rm -rf *
• This article uses language as a means to explain machine learning generalization with neat illustrations!
• Gaussian Processes sometimes seem to be on the fringes of modern machine learning, but they’re incredibly useful.

Got this from a friend? Subscribe here!

My Current Obsession

Something interesting happened. I noticed that the 3-week vacation was extremely relaxing and healthy for me. Between my PhD and the pandemic, I noticed not just physical decline (I gained 40kg and had some serious health stuff come up) but even cognitive decline. I noticed how I started forgetting people and situations from before I moved to Denmark. Whether it’s names or just nice things that happened. Mind you, I never had great memory, but it was concerning. I knew I was burnt out. But I am the sole provider for myself, if I don’t work, I don’t survive. My current job was already leagues better for my mental health, but mostly to the point that it wasn’t deteriorating or at least only slowly getting better. Then on my way home from the airport after 3 days of travel, I started remembering names of classmates in elementary school. I remembered what I considered core memories long ago. It was like something reset in my brain. I don’t say this miraculously cured my burnout, but it was a pause and finally some recovery from what has been a pretty bad situation for six years. I look forward to the coming time and what it will bring.

I’m also obsessed with the recent advances in stable diffusion. Considering creating a Skillshare course about them for artists and those who want to get started in prompt engineering. If you have a favourite resource about these, please let me know! You can just reply to this newsletter issue and it should get to me!

Also, scuba diving is the best. I want to do it more. Simple as that.

Thing I Like

This is an aspirational one. But I enjoyed diving so much that I want to get my own gear. I got a very kind response to my last recommendation of the Hollis mask to think about getting a regulator (the breathy thing), and I’ve been seriously looking at the Apeks XTX50. It’s so cool! Why do I like expensive hobbies again?

Hot off the Press

I published an article about testing in machine learning and another one about interpretability as tools to enable collaboration and make our scientific progress more streamlined and easy.

Data Stories

I love board games.

Monopoly isn’t my favourite unless I need a good argument in my friend group or family. But I love a good statistical analysis, which streets you should heavily invest in. This is from 100,000 games played:

Source: Reddit user AnalyticsReloaded

Question of the Week

• Why do we usually normalize the inputs to a neural network?>

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

• This Tweet by Sebastian Raschka about the efficiency of CNNs compared to MLPs.
• I love a good productivity app. I tried them all. But this little tool is new (and free!). Let your todos for a day battle it out in a todournament and see who rises to the top!
• What you can learn from the summary of 223 Y combinator startup pitches!