Online Semisupervised Learning Approach for Quality Monitoring of Complex Manufacturing Process

Algorithm 1

ParsNet++ algorithm.

	Input: partially labelled data batches:
	for data batch do
	Testing and update performance metrics
	if k < S then {S: initialization batch number}
	for epochs = 1:E do
	Update ACM
	{gen:generative phase}
	for alldo
	Structural evolution
	Calculate { in (1)}
	end for
	for alldo {dis:discriminative phase}
	Structural evolution
	Calculate { in (1)}
	end for
	end for
	else
	Update ACM
	if exists unlabelled data then
	Generate pseudolabel via (2)
	end if
	Calculate { in (1)}
	for alldo
	Structural evolution
	end for
	for alldo
	Structural evolution
	Calculate { in (1)}
	Update net with R in (1) {autoregularization}
	end for
	end if
	end for