算法思维举要

1. Everything is optimization. Problem solving = Representation->evaluation->optimization.
2. The solution is continuous or discrete ?
3. If discrete, could brute-force or branching be a choice ?
4. The environment has uncertainty ?
5. The problem can be divided and conquer ?
6. The problem can be solved recursively ?
7. Can we trade time(space) with space(time) ?
8. About artificial intelligence : 机器学习五大流派（https://cloud.tencent.com/developer/article/1053989）
   1. My Understanding:
      1. Symbolists : won't work in real life
      2. Connectionists: like inverse deduction by symbolists, but with BP
      3. Evolutionaries: heuristics. May be promising for problems with a structure, e.x. combinatorial problems
      4. Bayesians: master of probabilities. Useful in any real-life circumstances.
      5. Anologizers: never forget similarity. Well employed in unsupervised learning.

1. 递归
2. 分治
3. 对抗生成的思想：GAN, Actor-Critic...
4. 浓缩过滤思想：Embedding
5. 轮流迭代优化法：ADMM, EM, K-means等