iiinova — where your digital self keeps evolving

Will Coffey @willcoffey.bsky.social · 06/06/2025

www.kaggle.com/code/william... I struggled to find good sources on backpropagation and neural network implementations that were very light on background (and especially mathematical notation) knowledge. So I wrote one for others, just to expand the pile.

kaggle.com

Neural Networks For Dumb Idiot Babies

Explore and run machine learning code with Kaggle Notebooks | Using data from No attached data sources

000

Will Coffey @willcoffey.bsky.social · 06/01/2025

Advertisers, presumably. Not in marketing, but I suspect platform engagement is a primary driver for gauging advertising value. Bots are perceived as a way to keep people engaged even with activity down, or if they're engaging in a largely detestable way. It's a sad product of perverse incentives.

010

Will Coffey @willcoffey.bsky.social · 09/12/2024

Stumbled across this post on Substack by @deliprao.bsky.social today that I really appreciated as someone trying to break into the field. Simple categorizations can seem trite at times, but they can be deceptively profound in breaking down complex problems. substack.com/home/post/p-...

substack.com

Juicy Research Ideas and How to Find them?

How do people come up with research ideas in AI? Will the "AI Scientist" finally make me work full-time on my chicken farm?

211

Will Coffey @willcoffey.bsky.social · 28/11/2024

That'll inevitably change with time and audience dilution. But for what it's worth, I don't know if that's necessarily a bad thing. We've come to expect a mass-appeal homogenization of digital communities, but in theory it's probably healthier for there to be a diversity of social atmospheres.

000

Will Coffey @willcoffey.bsky.social · 17/10/2024

Imagine that for each of a school's programs, demographic "A" is accepted at a higher rate than demographic "B". Does that mean that "A" was accepted to the school at a higher rate than "B" overall? Nope! A brief and intuitive description of Simpson's Paradox, primarily to explain it to myself.

github.com

Notebook/Data Science/Simpson's Paradox/simpson.ipynb at master · WilliamCoffeyTri/Notebook

Contribute to WilliamCoffeyTri/Notebook development by creating an account on GitHub.

000