Ref: https://securityintelligence.com/factorization-machines-a-new-way-of-looking-at-machine-learning/

A new way of looking at machine learning

  1. Introduced by Steffen Rendle 2010 at Google.
  2. Can be compared to SVM with a polynomial kernel.
  3. SVM is kind of mystery(?)
  4. SVM works best on dense data.
  5. Factorization machines perform well on sparse data.
  6. Can model \(n\)-way variable interactions.
  7. Computational complexity can be reduced to linear.
  8. Optimization methods: stochastic gradient descent, alternating least-squares, Markov chain Monte Carlo (recommended), adaptive gradient descent.
  9. Widely used in collaborative recommendation systems.
  10. Recommend music, movies; predict stock market.
  11. libFM, fastFM, spark-libFM