Improve
- Add some neuralnetwork optimizers #863
- Add inverse function for transform and scaling models #865
- Add some neuralnetwork optimizer #871
- Add ELM #875
- Add DTSCAN #877
- Add DOC #879
- Add some iterative method of solve in Matrix class #881
- Add DiSH #885
Small changes
- Changed to use float16 instead of int16 #873
- Increase retry time of some tests #890
- Update version to 0.21.0 #891
Full Changelog: v0.20.1...v0.21.0