search

accosmin / nano

C++ library [machine learning & numerical optimization] - superseeded by libnano
MIT License
1 stars 0 forks source link

Support for other well-known machine learning datasets #154

Closed accosmin closed 7 years ago

accosmin commented 7 years ago

These datasets are small, well documented and easy to experiment with. Get the most downloaded once from: http://archive.ics.uci.edu/ml/

accosmin commented 7 years ago

Add task_t::describe(bool detailed) to output images or textual description of a task. And modify info_task accordingly.

May remove afterwards all the synthetic tasks (and experiments), the image warping (+ apps/embed) utilities. Also update README.

May add nano::dbdir() to return the default database directory for the supported datasets: ENV(HOME)/experiments/databases/. This should simplify the unit tests: have dir="" by default and if not given then nano::dbdir() will be used.