Given symmetric matrices B,D ∈ R n×n and a symmetric positive definite
matrix W ∈ R n×n , maximizingthe sum of the Rayleighquotientx ? Dx andthe gener-
alized Rayleigh quotient
x ? Bx
x ? Wx
on the unit sphere not only is of mathematical interest
in
关于Noisy Networks for Exploration dqn的原始论文，适合初学者对深度强化学习Noisy Networks for Exploration dqn的认识和了解Published as a conference paper at ICLR 2018
T is assessed by the action-value function Q defined as
Q"(.a)=配
∑
rR(t, at)
(1)
where E is the expectation ove
