New Year’s Resolutions 2022: Diverse and motivated

Posted on January 1, 2022

Because life is a never-ending learning process, every year I set myself new areas to improve. For many years I have been writing publicly my new year resolutions (2021,2020, 2019, etc) and my own assessments at the end of the year (2021,2020, 2019, etc). The process of writing them and after 365 days revisiting them is a practice that I find useful. This time I won’t split the objectives into different categories. Instead I’ll merge them together wihtout any lengthRead More→

2021 resolutions revised: PhD done, nothing else though

Posted on December 30, 2021

In Denmark, we started 2021 deep in lockdown. Things gradually improved early summer until we got an almost normal life back in December. This year was my last PhD year where I also submited my thesis (still need to defend). After reviewing the goals I can sum it up as lots of red. Personal Physical exercises: I kept doing regular workouts, specially runs but also bike rides and swims. This year I did my personal best times in semi-official races.Read More→

Stop explaining black box machine learning models for high stakes decisions and use interpretable models instead (paper summary)

Posted on December 7, 2021

Deep learning models are usually regarded as black boxes. That is because they are not transparent about the way they reach the prediction. Humans cannot directly interpret the model with millions of parameters. Choosing ignorance can lead to unforeseen dangers. This is inherently a bad practice that should be minimized as much as possible. Current deep learning explainability tools aim to simplify the process to the outcome but does not really explain the “thinking process” that the model followed. ThisRead More→

Hidden technical debt in machine learning systems (paper summary)

Posted on November 30, 2021

Machine learning systems are wonderful. Many shapes and forms of machine learning algorithms are currently in use. Different models such as clustering like k-means, prediction methods like trees, or more advanced deep learning methods suffer from technical debt. In traditional software engineering, technical debt can be found in specific shapes. In addition to the “traditional” software engineering problems, machine learning systems also face new challenges. The following paragraphs present the different technical debt found in machine learning systems. 1. EncapsulationRead More→

Gain specific knowledge by making an observation

Posted on November 15, 2021

As defined by Naval Ravikant, specific Knowledge is the knowledge that you cannot be trained for it. It is located at the edge of knowledge and is very hard to figure out. To acquire specific knowledge requires a combination of existing tools but also creativity to combine them in innovative ways to create something new. To gain specific knowledge in your field you need to experiment. Everything starts with the question “what would happen if?“. 1. Make an observation Let’sRead More→

How to build assets?

Posted on November 2, 2021

Generally, when people talk about investing they think it’s about buying assets. However, that’s not the case. There are plenty of activities in which one can build assets with little to no money. If you create a blog, you’re creating an asset. If you learn, you’re improving your asset. When you go to work, you’re creating value. All of these activities have different pros and cons but one could say that all of them will bring you more money. SomeRead More→

Moving from power to talent

Posted on November 1, 2021

This is the most profound shift of all. Moving from the J.P. Morgan model of ambition to the Mark Zuckerberg model shifts the balance of power from capital to talent. Ambitious people have gone from writing cheques to writing code. Today the most ambitious individuals don’t own the means of production, if they can write code they are their own means of production (Marx would, perhaps, be surprised). This gives ambitious people unprecedented power. https://medium.com/entrepreneur-first/tech-entrepreneurship-and-the-disruption-of-ambition-4e6854121992

Worth mentioning III: Acquisition loops, NFTs explained, Orson Welles and friendship, and B2B sales

Posted on July 20, 2021

This is the continuation of the worth mentioning experiment where I shared relevant links to be kept for future references. Acquisition Loops: The old ways of acquiring new customers -aka funnels- are outdated. Modern brands are adapting their marketing strategy and shifting their tactics to acquisition loops. In this article, there is an explanation accompanied by some examples from relevant brands. NFTs, explained: Currently cryptocurrencies are in everybody’s mouth. Even newer trends are the Non-Fungible Tokens (NTF) which are aRead More→

The Veblenian Entrepreneur

Posted on May 30, 2021

For quite a long time I’ve been thinking that people tend to focus their time on easy things instead of focusing on the complicated ones in an attempt to avoid pain. People, in general, try to avoid harmful situations which leads them to take the path with the least pain possible (at least at the start) i.e. the easy path. In some cases, the easy and straightforward choice may become the worse long-term solution. But this scientific publication made meRead More→

Roulette Wheel Selection in Python 🐍

Posted on March 22, 2021

The roulette wheel selection (also known as fitness proportionate selection) is a function used by genetic algorithms for selecting potentially useful solutions for recombination. The crossover individual probability is computed based on the individual’s fitness divided by the sum of all population fitness. The following is the formula for it: where pi is the probability of each chromosome equals the chromosome frequency divided by the sum of all fitness. Let’s imagine that the roulette wheel selection algorithm is like aRead More→