Complete result settings for Audio dataset (table columns are sortable)
The experiments are described in Rooshenas and Lowd, Discriminative Structure Learning of Arithmetic Circuits, AIStats 16
| Tree# | C | Max. EM Iteration | Train CLL | Validation CLL | Test CLL | Learning Time |
| 1 | 0.010000 | 20 | -19.287970 | -19.322679 | -19.362315 | 1610.221317 |
| 1 | 0.050000 | 20 | -18.781706 | -18.990117 | -18.980283 | 1582.653574 |
| 1 | 0.100000 | 20 | -18.734434 | -18.968612 | -18.964714 | 1611.303349 |
| 1 | 0.500000 | 20 | -18.719312 | -18.980860 | -18.984955 | 1631.223073 |
| 1 | 2.000000 | 20 | -18.718407 | -18.986954 | -18.991577 | 1814.938966 |
| 1 | 1.000000 | 20 | -18.718597 | -18.984716 | -18.989187 | 1864.641492 |
| 2 | 0.100000 | 10 | -18.496484 | -19.068195 | -19.083248 | 24619.024570 |
| 2 | 0.500000 | 10 | -18.414362 | -19.139552 | -19.188382 | 26764.442228 |
| 2 | 2.000000 | 10 | -18.405323 | -19.178529 | -19.191323 | 27111.761470 |
| 2 | 1.000000 | 10 | -18.404648 | -19.188616 | -19.196443 | 27126.823950 |
| 3 | 2.000000 | 10 | -18.090605 | -19.437055 | -19.438042 | 49861.021329 |
| 3 | 0.100000 | 10 | -18.283472 | -19.171922 | -19.188772 | 43059.468004 |
| 3 | 1.000000 | 10 | -18.101214 | -19.443297 | -19.461995 | 49015.602238 |
| 3 | 0.500000 | 10 | -18.126064 | -19.388121 | -19.389006 | 47720.498398 |
| 4 | 2.000000 | 10 | -17.790379 | -19.646503 | -19.737065 | 78537.239239 |
| 4 | 0.100000 | 10 | -18.085136 | -19.261039 | -19.270451 | 63978.732614 |
| 4 | 1.000000 | 10 | -17.799380 | -19.654569 | -19.686362 | 76436.149593 |
| 4 | 0.500000 | 10 | -17.858230 | -19.547462 | -19.623949 | 74873.659592 |