A Data Mining Method Using Deep Learning for Anomaly Detection in Cloud Computing Environment

Algorithm 2

Compute

	input data set
	k: number of nearest neighbor
	: threshold for LOF
	N: data block number
	output data set which
	Initialize a Hadoop Job
	Set TaskMapReduce class
	Logically divide X into multiple data blocks: .
	In the -th TaskMapReduce
	FirstMapper
	input
	output <key, value> = < >
	for each data di, i = 1, 2, ..., m do
	Calculate disij = distance (di, dj), j = 1, ... , m
	Sort disij of di
	for each disij of di do
	if &
	add di and disij in k-distinct-neighbor record (, )
	end
	Calculate k-distinct-distance record k-dis (di)
	end
	First Record
	input <key, value> = < di, [(ok, dis (di, ok)), k-dis (di) >
	output <key, value> = < di, [(ok, dis (di, ok)), k-dis (di) >
	SecondMapper
	input < key, value> = < di, [(ok, dis (di, ok)), k-dis (di) >
	output < key, value> = < di, [(ok, reach-dis (di, ok)) >
	for ok ∈ k-distinct-neighbor do
	if k-dis (di) < dis (di, ok)
	reach-dis (di, ok) = dis (di, ok)
	else reach-dis (di, ok) = k-dis (di, ok)
	end
	SecondReducer
	input < key, value> = < di, (ok, reach-dis (di, ok)) >
	output < key, value> = < di, lrd (di) >
	for value do
	,
	ok ∈ k-distinct-neighbor
	end
	ThirdMapper
	input < key, value> = < di, lrd (di) >
	output < key, value> = < di (lof (di) > θ), lof (di) >
	for ok ∈ k-distinct-neighbor do
	,
	ok ∈ k-distinct-neighbor
	end
	if lof (di) > θ
	output
	ThirdReduce
	input < key, value> = < di (lof (di) > θ), lof(di) >
	output < key, value> = < , lof () >
	for value do
	Sort for and record
	End