Revision 0f9ce693bda1e6591173af4de60a2cfa610f4957 authored by Marie Roald on 06 June 2021, 08:48:42 UTC, committed by Marie Roald on 06 June 2021, 08:48:42 UTC
Only making the constrained factor matrices non-negative seemed to make the fitting procedure less stable (higher likelihood of all-zero components).
1 parent 5f08046
| File | Mode | Size |
|---|---|---|
| backend | ||
| contrib | ||
| datasets | ||
| decomposition | ||
| metrics | ||
| random | ||
| regression | ||
| tenalg | ||
| tests | ||
| utils | ||
| __init__.py | -rw-r--r-- | 2.1 KB |
| _factorized_tensor.py | -rw-r--r-- | 716 bytes |
| base.py | -rw-r--r-- | 5.5 KB |
| cp_tensor.py | -rw-r--r-- | 20.7 KB |
| parafac2_tensor.py | -rw-r--r-- | 19.0 KB |
| testing.py | -rw-r--r-- | 837 bytes |
| tt_matrix.py | -rw-r--r-- | 8.3 KB |
| tt_tensor.py | -rw-r--r-- | 10.2 KB |
| tucker_tensor.py | -rw-r--r-- | 13.6 KB |
Computing file changes ...
