A Few Useful Things to Know about Machine Learning - Githubissues

ssirai / reading

0 stars 0 forks source link

A Few Useful Things to Know about Machine Learning #13

Closed ssirai closed 9 years ago

ssirai commented 9 years ago

dga.ps - cacm12.pdf

ssirai commented 9 years ago

[x] INTRODUCTION
[x] LEARNING=REPRESENTATION + EVALUATION + OPTIMIZATION
[x] IT’S GENERALIZATION THAT COUNTS
[x] DATA ALONE IS NOT ENOUGH
[x] OVERFITTING HAS MANY FACES
[x] INTUITION FAILS IN HIGH DIMENSIONS
[x] THEORETICAL GUARANTEES ARE NOT WHAT THEY SEEM
[x] FEATURE ENGINEERING IS THE KEY
[x] MORE DATA BEATS A CLEVERER ALGORITHM
[x] LEARN MANY MODELS, NOT JUST ONE
[x] SIMPLICITY DOES NOT IMPLY ACCURACY
[x] REPRESENTABLE DOES NOT IMPLY LEARNABLE
[x] CORRELATION DOES NOT IMPLY CAUSATION
[x] CONCLUSION
[ ] REFERENCES