Advanced Search

Browse by Discipline

Scientific Societies

E-print Alerts

Add E-prints

E-print Network

  Advanced Search  

Learning a Circuit by Injecting Values [Extended Abstract]

Summary: Learning a Circuit by Injecting Values
[Extended Abstract]
Dana Angluin
James Aspnes
Jiang Chen
Yinghua Wu
We propose a new model for exact learning of acyclic cir-
cuits using experiments in which chosen values may be as-
signed to an arbitrary subset of wires internal to the circuit,
but only the value of the circuit's single output wire may
be observed. We give polynomial time algorithms to learn
(1) arbitrary circuits with logarithmic depth and constant
fan-in and (2) Boolean circuits of constant depth and un-
bounded fan-in over AND, OR, and NOT gates. Thus, both
AC0 and NC1 circuits are learnable in polynomial time in
this model. Negative results show that some restrictions
on depth, fan-in and gate types are necessary: exponen-
tially many experiments are required to learn AND/OR cir-
cuits of unbounded depth and fan-in; it is NP-hard to learn


Source: Aspnes, James - Department of Computer Science, Yale University


Collections: Computer Technologies and Information Sciences