1、Principal Component Pursuit(PCP)
1.1 PCP:
1.2 Iterative optimization methods:
Accelerated Proximal Gradient(APG)
Augmented Lagrangian Multiplier(ALM)
Difficulty applied to Online: nuclear norm tightly couples the samples and thus the samples have to be processed simultaneously.
2、OR-PCA
An equivalent form of the nuclear norm:
where rank
(X)
is upper bounded by
r
.
R∈Rn×r
: the coefficients of the samples w.r.t. the basis.
Equivalent to minize empirical cost function:
where the loss function for each sample is defined as:
Expected cost over all the samples:
OR-PCA algorithm: Develop a stochastic optimization algorithm to processes one sample per time instance.
minimize the empirical cost function to update
r
, and
e
:
where ℓ(zi,L) is:
Optimize a surrogate function(with block-coordinate descent) of
ft(L)
to update the basis
Lt
:
which provides an upper bound for ft(L):gt(L)≤ft(L) .
3、模型求解
In the t-th time instance, minimizing the cumulative loss w.r.t. the previously estimated {ri}ti=1 and {ei}ti=1 to estimate the basis Lt .