| … | … | @@ -94,6 +94,11 @@ Do we need to have explicit "must-form-a-cycle" rules? Can we do this |
| 94 | 94 | statistically? |
| 95 | 95 | |
| 96 | 96 | |
| 97 | ### Random walks through possibilities | |
| 98 | Walk through possible connections, randomly. But rather than uniform | |
| 99 | weighting, we need to spread "attention" according to weights. This | |
| 100 | would require a redesign of the attention allocation subsystem. | |
| 101 | ||
| 97 | 102 | |
| 98 | 103 | Literature |
| 99 | 104 | ========== |