Mercurial > pylearn
comparison doc/v2_planning/dataset.txt @ 1338:91637815b7ca
Added a comment on the dataset vs. task issue
| author | Olivier Delalleau <delallea@iro> |
|---|---|
| date | Thu, 21 Oct 2010 14:03:12 -0400 |
| parents | 7dfc3d3052ea |
| children | 158493f8dff9 |
comparison
equal
deleted
inserted
replaced
| 1337:7dfc3d3052ea | 1338:91637815b7ca |
|---|---|
| 565 - It is also ok to have datasets that do not support random access (so the | 565 - It is also ok to have datasets that do not support random access (so the |
| 566 only way to access samples is through iteration). | 566 only way to access samples is through iteration). |
| 567 - Ideally, data should be deterministic (i.e. __call__() should always | 567 - Ideally, data should be deterministic (i.e. __call__() should always |
| 568 return the same thing). It would probably be up to the user to be super | 568 return the same thing). It would probably be up to the user to be super |
| 569 careful if he decides to use a non-deterministic dataset. | 569 careful if he decides to use a non-deterministic dataset. |
| 570 | 570 - About the "task vs. dataset" distinction. This could be achieved by |
| 571 associating to a task the names of the fields it requires (e.g. "input" | |
| 572 and "target" for the regression task), and if the dataset does not | |
| 573 already defines these fields, using a dataset wrapper than does it | |
| 574 (saying for instance that "input" is the concatenation of "x1" and "x2", | |
| 575 and "target" is "y", for a dataset whose fields are x1, x2 and y). | |
| 576 |