AI
From Titanic to EEG Decoding
The unlikely thread from a first Kaggle notebook to reading neural signals — same habits, different substrate.
7 minMLEEGundergraduate
Same habits, different substrate
My first serious machine learning project was the Titanic notebook on Kaggle. It felt small at the time — a toy dataset, a simple classifier. Two years later I was training a decoder on EEG recordings, and I realized the habits were the same.
What carried over
- Start with a stupid baseline. Beat it before doing anything clever.
- Look at the data. Actually look at it, one row at a time.
- Do not trust a metric you did not compute yourself at least once.
What is different
EEG has time. It has noise you cannot preprocess away. It has subjects that behave nothing like each other. The Titanic notebook did not prepare me for that — but the discipline did.
The takeaway
Small projects, finished honestly, compound. The subject changes, the craft does not.