| 摘要 |
PURPOSE: To provide a learning device for a recurrent neural network which learns the recurrent network fast with high precision by using a second-order derivative. CONSTITUTION: An initialization part 21 initialize coupling weightΦ<1> and control variables H<1> and (k), and a stop condition decision part 22 decides the stop conditions of algorithm. When the stop conditions are not met, a correction direction calculation part 23 calculates a correction direction vectorΔΦ<k> =-H<k> ...f(Φ), an optimum search step width calculation part 24 calculates optimum step widthλ<k> minimizing an object function f(Φ<k> +λ<k>ΔΦ<k> ), and a coupling weight update part 25 updates the coupling weightΦ<k+1> =Φ<k> +λ<k>ΔΦ<k> . An approximate matrix calculation part 26 calculates H<k+1> on the basis of the conditions and the stop condition decision part 22 decides the stop conditions.
|
