AIDan Luuover 12 years ago
PCA is not a panacea
5 min read
A tech startup interviewer insisted PCA solves any classification problem. The candidate argued matrix factorization would be a better starting point, citing LSA's limits in capturing only pairwise word correlations versus deep learning. The disagreement consumed half the interview time.
Level
Hype check
Tap to vote and see what everyone thinks.