Learning a regular language by inferring a DFA with the TTT algorithm
8 hours ago
- #TTT_algorithm
- #automata_learning
- #regular_languages
- TTT algorithm combines discrimination trees and binary search for efficient regular language learning.
- It eliminates redundant membership queries by adding only one discriminator per counterexample.
- The algorithm uses prefix transformation for minimal access sequences and discriminator finalization for shallow trees.
- Comparison with L* shows TTT avoids redundant queries and uses a discrimination tree instead of a flat table.
- Examples include learning languages like strings with even a's, ending in 'b', or binary strings divisible by 3.