6Probe

6Probe assembles an Address Space Forest by repeating Divide Hierarchical Clustering with different splitting heuristics, converting each stable region into an address pattern, and sampling those patterns while preserving low-entropy structure.

• Reference: Placeholder citation (coming soon).

Train

rmap train --seeds seeds/hitlist.txt --output models/6probe.bin six-probe \
  --beta 12 --tree-num 40

Generate

rmap generate --model models/6probe.bin --count 250000 \
  --unique --output 6probe.txt

Configuration

• --beta <usize> – minimum region size fed into each DHC iteration (default 12).
• --tree-num <usize> – number of additional random forest trees (default 40).

Model notes

• Increasing --tree-num adds more randomized strategies, improving coverage at the cost of training time.
• The generator records wildcard positions in each pattern so that randomization only touches the entropy-heavy nybbles.