That's Fresh! Newsletter

FROM THE AI WORLD

I've recently stumbled into this nicely curated repository containing a list of machine learning projects that failed miserably. This list includes different model families and domains and links to relevant news sources. You might have heard about these ML incidents individually while reading the news; however, having a repository collecting and updating them is an excellent idea. I was familiar with the most famous ones, such as the Apple Credit Card or Zillow projects; nevertheless, I was happy to learn about less-known ones.

ML project failures can arise for different reasons, ranging from poor problem definition to problems with the data used to train the model. The data aspect will become more prevalent as large-scale NLP and computer vision models use massive datasets for training. These datasets are becoming increasingly difficult to clean and curate, so it is equally challenging to control what models learn. As they say, garbage in, garbage out!

ML projects epic fails

Despite the potential, ML projects may fail. In this Github library curated by Kenneth Leungty you'll find real world examples from which we can learn a lot.

INTERVIEW

How to build a data strategy

In one of our latest episodes of "People Also Ask...and we answer!" we interviewed Alberto Danese, Head of Data at Nexi, and discussed about data strategies.

HOT FROM THE PRESS

Clearbox AI and BearingPoint

We recently started a partnership with BearingPoint, leader in the technology consulting, for fraud detection and help fintech institutions. (Lang: 🇮🇹)

WEEKLY MEME

Your pals may also like...