Sonja Georgievska

9 posts

Research Software Engineer exploring explainable AI, machine learning, and data science.

Deep Learning6 Neural Networks5 Machine Learning4 Training3 RSE2 Collaboration1 Python1 Health1

Feb 23, 2023

The hidden dangers of using XAI carelessly

Congrats! Your ML model is finally XAI-ready. But before you start patting yourself on the back, have you pre-processed your features? If so, watch out for those SHAP values, they might have been “pro...

Dec 16, 2022

Small-scale initiative in Machine Learning 2021: how did it go?

Photo by "My Life Through A Lens" on Unsplash Small-scale initiative in Machine Learning 2021: how did it go? Introducing a series of blog posts on our collaborative projects in machine learning Here...

May 13, 2022

Reusable software in the era of AI: why sometimes you must reinvent the wheel

> In traditional software development, reinventing the wheel is a cardinal sin. But AI has changed the picture. In AI-based software development the traditional reusable components, like data aggregat...

Jun 18, 2020

Machine learning: when it is easy & when it is difficult

Hey, should I draw a wild curve that splits all points exactly? But how am I sure that the curve will split properly all new points that I haven’t seen, that it will not over-fit on the training data?...

Jun 18, 2020

Machine learning when it is easy & when it is difficult

A crash course in machine learning; only high-school level of math is required before reading. Photo by Element5 Digital on Unsplash Machine learning (ML) is about having the computer learn how to con...

Apr 25, 2020

Forcing science with brute force deep learning

When we are children, we lack patience. When we grow up, “we lack time”. Not all is bad, of course. Once trained, the deep neural networks are very fast when used for prediction. However, that assumes...

eScience

Apr 25, 2020

Why AI will not replace experts

Let’s take a few steps back. It is 1990 and a team of medical researchers equipped with a statistician is performing a randomized control study91656-U/fulltext) to find out if phenomenon X causes dise...

Jun 20, 2019

turning a panda into a cat!

Why adversarial examples are not scary…but are most likely useful “Is it a bird? Is it a plane?” A child who never saw Superman might mistake him for a bird or a plane. On the other hand, who knows wh...

Jun 12, 2019

explainable ai exit or infinite loop

> Let us simplify before we try to explain Photo by Franck V. on Unsplash Before we start, let’s make it clear: by AI here I mean computational statistics, and by explainable — well, I don’t really kn...

← All posts