Keeping Up With Data — Week 29 Reading List
5 minutes for 5 hours’ worth of reading
The image above comes from an article about research at Google in 2020, specifically about AutoML. It shows an evolutionary approach to meta-learning (using learning algorithms to develop new ML algorithms). The Auto-ML Zero can learn algorithms developed in the last thirty year. Is this the approach that will win the race towards the artificial general intelligence?
Ok, let’s not jump the gun! Data science tools in the browser, trends in ML and data-driven thought leadership are on the menu this week.
- JupyterLite: Jupyter ❤️ WebAssembly ❤️ Python: I came across JupyterLite just recently. It’s a JupyterLab distribution that runs entirely in the browser. Its advantage is that it can be deployed as a static website. Without any complicated server setup. Just a simple HTTP server to serve the static files to users. Let’s see how it will land with the community. JupyterLite doesn’t have all the extensions available in JupyterLab (yet), but it’s already looking great. Check it out here: https://jupyterlite.github.io/demo/lab/ (Jeremy Tuloup @ jupyter)
- Trends in Machine Learning Theory: We’ve seen a boom of machine learning in the recent years. Both, when it comes to progress in the theoretical research as well as its real-life application. What are the trends in the learning theory? New learning paradigms (a quest to develop more realistic models of learning), trustworthy machine learning (focusing on privacy, robustness, fairness, interpretability, and causality), and reinforcement learning (can an agent learn from exploring a finite number of states and generalise to perform well on unknown areas of the environment?). (The Simon’s Institute Blog)
- Data-Driven Isn’t Dead: Some voices are saying that “data-driven” should be thrown out of the vocabulary and companies should be data-informed or data-conscious. Tomato tomahto! It’s the meaning that matters. Being data-driven is about leveraging data to drive optimal outcomes. The article argues that the issue is in how the data is being used. Predictive analytics can lead to locally optimal outcomes. Prescriptive analytics is needed to reach the global optimum. (Fabrizio Fantini @ tds)
The data stack has been changing so fast that one often feels that the only certainty is SQL. And then you come across an article going against SQL! Luckily (my opinion), SQL is still alive and kicking. It has an incredibly steep learning curve and one can get a lot done very quickly (to master it, of course, takes years). And it seems, that I’m not the only one, who’s for SQL.