Properly Learning Decision Trees in almost Polynomial Time
- Stanford, Stanford, CA, USA; OSTI
- MIT, Cambridge, MA, USA
- Stanford, Stanford, CA, USA
We give annO(log logn)-time membership query algorithm for properly and agnostically learning decision trees under the uniform distribution over { ± 1}n. Even in the realizable setting, the previous fastest runtime wasnO(logn), a consequence of a classic algorithm of Ehrenfeucht and Haussler.
Our algorithm shares similarities with practical heuristics for learning decision trees, which we augment with additional ideas to circumvent known lower bounds against these heuristics. To analyze our algorithm, we prove a new structural result for decision trees that strengthens a theorem of O’Donnell, Saks, Schramm, and Servedio. While the OSSS theorem says that every decision tree has an influential variable, we show how every decision tree can be “pruned” so thateveryvariable in the resulting tree is influential.
- Research Organization:
- Stanford Univ., CA (United States)
- Sponsoring Organization:
- USDOE Office of Science (SC)
- DOE Contract Number:
- SC0019205
- OSTI ID:
- 2421027
- Journal Information:
- Journal of the Association for Computing Machinery, Journal Name: Journal of the Association for Computing Machinery Journal Issue: 6 Vol. 69; ISSN 0004-5411
- Publisher:
- Association for Computing Machinery (ACM)
- Country of Publication:
- United States
- Language:
- English
Similar Records
Polynomial time approximation schemes for Euclidean TSP and other geometric problems
A polynomial time primal network simplex algorithm for minimum cost flows