Moustakides, G.V., "Study of the Transient Phase of the Forgetting Factor RLS," IEEE Trans. on Signal Processing, Vol. 45, No. 10, pp. 2468-2476, Oct. 1997.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....sensitivity analysis (see, e.g. HP00] transient stability prediction (see, e.g. LT00] and trajectory tracking (see, e.g. TVSS98] might help to find solutions for the transient management problem, as well. Convergence of learning algorithms: There are many papers (e.g. CMS99] [Mou97], MA98] Solo97] and [Sun93] that investigate the stability and convergence properties of adaptive algorithms. The motion of adaptive systems from their initial states to their final (stationary) state can be considered as a transient motion, therefore every method for the characterization of ....

Moustakides, G.V., "Study of the Transient Phase of the Forgetting Factor RLS," IEEE Trans. on Signal Processing, Vol. 45, No. 10, pp. 2468-2476, Oct. 1997.

....and McWhirter. In further work it would be interesting to investigate in more detail the stability, numerical complexity and convergence properties of IPM as compared to QRD RLS. Another issue is robustness to initialization. It is now understood that RLS is quite sensitive to its initialization [7]; IPM however, can be initialized almost arbitrarily at least in the unconstrained case. ....

G. V. Moustakides, "Study of the transient phase of the forgetting factor RLS," IEEE Transactions on Signal Processing, vol. 45,

....it determines the exact amount of regularization for the remainder of the adaptive process. In the absence of estimation uncertainities about R xx (n) and p xy (n) it is the regulatization term that determines the rate at which the RLS estimator converges to the optimum linear filter (see also [5]) 1 In this paper we reformulate the adaptive filtering problem as a convex feasibility problem and propose to use the analytic center (cf. Section 2) of the feasible region as an estimate of the optimal filter R xx (n) Gamma1 p xy (n) This estimator takes the form (approximately) of (ff n ....

....out the effect of initialization and is therefore more algorithm specific. We shall refer to this algorithm property as its transient behaviour. This is consistent with other work in which the initialization of RLS has been identified as the most dominant factor governing transient convergence [5]. Fast transient convergence is important for accurate estimation of time varying parameters. The two types of convergence behaviour can be nicely separated as follows. To isolate the statistical averaging process, we assume that the algorithm is initialized in such a way that there is no need for ....

[Article contains additional citation context not shown here]

G. V. Moustakides, "Study of the transient phase of the forgetting factor RLS," IEEE Transactions on Signal Processing, vol. 45, pp. 2468--2476, October 1997.

....out the effect of initialization and is therefore more algorithm specific. We shall refer to this algorithm property as its transient behaviour. This is consistent with other work in which the initialization of RLS has been identified as the most dominant factor governing transient convergence [10]. The above mentioned two types of convergence behaviour can be nicely separated as follows. To isolate the statistical averaging process, we can assume that the algorithm is initialized in such a way that there is no need for the phasing out of effects of initialization. On the other hand, to ....

G. V. Moustakides, "Study of the transient phase of the forgetting factor RLS," IEEE Transactions on Signal Processing, vol. 45, pp. 2468--2476, October 1997.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC