Algorithm 1: Malware Classification Using Transfer Learning and Texture Features
Input	PCAP
Output	Malware Classification
Step 1:	Set P= {${\mathrm{p}}_1$, ${\mathrm{p}}_2$, …, ${\mathrm{p}}_{\mathrm{n}}$}s, where is P is a Packets
Step 2:	$\mathrm{Decrypt} \left(\mathrm{P}\right)= \mathrm{P}\prime$
	Compute PCAP from $\mathrm{P}\prime$, where $\mathrm{P}\prime$ $=\left(\mathrm{IP}, \mathrm{TCP},\mathrm{HTTP},\dots ,\mathrm{n}\right)$
	Selects NF from PCAP, where NF is the required Network Flows
	Display/Select HTTP + TCP
Step 3:	Select HTTP traces and TCP flows
Step 4:	Tokenize and filter HTTP and TCP flows = Clean features
Step 5:	Apply word embedding
	BERT transformers = Train feature
	Extraction = Textual trained features
Step 6:	Trained files = Trained features as $\mathrm{T}$
Step 7:	Compute B $=\left\{{\mathrm{B}}_1,{ \mathrm{B}}_2, \dots .,{ \mathrm{B}}_{\mathrm{n}} \right\}$ from PCAP, where $\mathrm{B}$ for Bytes
	Compute $\mathrm{I}$, where $\mathrm{I}$ is Image
	Decomposed $\mathrm{I}$ in ${\mathrm{SS}}_1$, where ${\mathrm{SS}}_1=256 \times 256$
	Apply FAST extractor & BRIEF descriptor on ${\mathrm{SS}}_1$
Step 8:	Generate texture features from the combination of FAST and BRIEF
Step 9:	Get Texture features as $\mathrm{T}\prime$
Step 10:	Combine $\mathrm{T}, \mathrm{T}\prime$ (Textual and texture features)
Step 11:	Apply SMOTE classing balancing on $\mathrm{T}$, $\mathrm{T}\prime$
	Compute $\mathrm{BT}$, ${\mathrm{BT}}^{\prime }$ from $\mathrm{T}$, $\mathrm{T}\prime$, where $\mathrm{BT}$, ${\mathrm{BT}}^{\prime }$ are Balanced Textural and Texture features
	$\mathrm{BT}=\mathrm{CNN}\left(\mathrm{T}\right),$ to apply CNN technique of trained features
	${\mathrm{BT}}^{\prime }=\mathrm{CNN}\left( \mathrm{T}\prime \right)$, to apply CNN technique of texture features
Step 12:	Calculate Deep Features as DF
Step 13:	Apply Voting-based ensemble learning on DF
Step 14:	Result: Malware or Benign
Step 15:	Finish