model performance 0.002051594
initial model 0.001913297
model selection 0.001851607
popular model 0.001814867
ferred model 0.001814867
bayesian word 0.001749166
word segmentation 0.001632808
word proportions 0.001611435
model 0.0016074
times word 0.001606866
training data 0.001371243
additional data 0.0012848220000000001
new data 0.001217986
test set 0.0010938179999999999
first set 0.001087768
same particle 0.001016029
training sequence 0.001014113
new set 9.989629999999998E-4
probability distribution 9.65428E-4
particle weight 9.569490000000001E-4
different initialization 9.47084E-4
other tokens 9.41489E-4
different gibbs 9.389769999999999E-4
particle algorithm 9.078090000000001E-4
bayesian models 9.074549999999999E-4
learning algorithm 8.97877E-4
initialization set 8.97339E-4
different reservoir 8.96586E-4
other factors 8.95643E-4
ing storage 8.68795E-4
state space 8.58129E-4
same constraints 8.45495E-4
new state 8.43915E-4
particle weights 8.38885E-4
ing tokens 8.37605E-4
ing techniques 8.33351E-4
sampling algorithms 8.31679E-4
initial models 8.29066E-4
gibbs sampling 8.13116E-4
ple set 8.097509999999999E-4
online algorithm 8.020659999999999E-4
topic models 7.97893E-4
ing approximation 7.96721E-4
importance sampling 7.85915E-4
previous experiments 7.85449E-4
previous research 7.79505E-4
good performance 7.77523E-4
proposal distribution 7.775E-4
dynamic state 7.76979E-4
stop words 7.76226E-4
rejuvenation sequence 7.721290000000001E-4
reservoir sampling 7.70725E-4
sampling reservoir 7.70725E-4
posterior distribution 7.50148E-4
sampling history 7.48993E-4
transition probability 7.450180000000001E-4
long sequence 7.37984E-4
particle filter 7.361760000000001E-4
online learning 7.35529E-4
effects sampling 7.34269E-4
overall algorithm 7.30938E-4
ing split 7.28934E-4
state transitions 7.27718E-4
equal probability 7.27279E-4
probability masses 7.22021E-4
previous states 7.162620000000001E-4
linear storage 7.14455E-4
english stop 6.98865E-4
random initial 6.82738E-4
state transition 6.783E-4
importance weights 6.733570000000001E-4
juvenation sequence 6.72109E-4
new states 6.61783E-4
eral methods 6.60328E-4
cle state 6.580550000000001E-4
state tran 6.575999999999999E-4
bayesian inference 6.5642E-4
sample size 6.55855E-4
common example 6.55727E-4
current work 6.53442E-4
initialization sample 6.51942E-4
rejuvenation step 6.48147E-4
particle filters 6.407100000000001E-4
filters particle 6.407100000000001E-4
mean nmi 6.40137E-4
particle cloud 6.37679E-4
latent dirichlet 6.37368E-4
benefits particle 6.37333E-4
particle fil 6.34649E-4
constant space 6.27776E-4
tion step 6.26284E-4
streaming learning 6.2439E-4
new observation 6.23678E-4
initialize weights 6.22325E-4
future work 6.20289E-4
space asymptoti 6.19781E-4
good sam 6.09102E-4
large rejuvenation 6.08027E-4
sample sizes 6.0396E-4
storage complexity 6.01662E-4
