Use machine learning methods for chemometric data from mass spectrometers.
This new modality of data corresponds to human smell and taste. We already have sensors for sound, images, location, movements, temperature, pressure, etc. This could be a "digital nose".
It could be used to:
- classify samples
- compute similarity between two samples
- find similar samples for a given one
- visualize a set of samples in 2D
The kinds of samples could be eg.:
- food
- drinks
- beer
- wine
- whiskey
- parfumes
- smells of
- flowers
- animals
- people
- dangerous chemicals
- ordinary industrial chemicals
- quality of water
- quality of chimneys or other heat-sources
- contamination of soil with chemicals
- landmine detection
- composition of materials
- metals
- precious stones
- etc.
Usable hardware: Scio project (see Kickstarter).