search

cognoma / machine-learning

Machine learning for Project Cognoma
Other
32 stars 47 forks source link

Isomap #78

Closed KT12 closed 7 years ago

KT12 commented 7 years ago

Used isomap to run non-linear dimensionality reduction before using logistic regression. Because of the computing resources needed, the nearest neighbors used were 8 and the components retained were 64. I ran it on a machine with 8GB RAM and 8GB Ubuntu swap.

The performance of the classifier would improve with higher settings, but would be difficult to execute on my machine.

dhimmel commented 7 years ago

The performance of the classifier would improve with higher settings, but would be difficult to execute on my machine.

It's certainly not performing that well with the current paramenters (: Good to know.

KT12 commented 7 years ago

Thank you for merging.

At least we now know that the gene expression features do not lie on a curved manifold and linear dimensionality reduction will get good, computing resource-efficient classification.