• Rajanbabu, T
• SreyaChandran
• J.D. Swethavarshni

This paper introduces CapsuleLSTM, a cutting-edge framework that leverages the power of Capsule Networks (CapsNets) and Long Short-Term Memory (LSTM) — a type of recurrent neural network (RNN) — for robust image spam detection. The objective is to categorize images into spam and ham, addressing the challenges posed by unwanted material presented in image form. CapsNets bring a fresh perspective to feature learning by capturing hierarchical relationships within images, while LSTM excels in modeling sequential dependencies crucial for understanding nuanced patterns. The p roposed CapsuleLSTM methodology undergoes rigorous evaluation using standard test datasets, including Dredze Dataset, Image Spam Hunter (ISH) Dataset, and Improved Dataset. To mitigate data scarcity, transfer learning and data augmentation techniques are integrated. The fully connected (FC) layer in CapsNets is seamlessly fused with LSTM, ensuring an effective combination of hierarchical feature learning and sequential context preservation. CapsuleLSTM represents a significant advancement in image spam detection, synergizing the strengths of CapsNets and LSTM to deliver a powerful, efficient, and nuanced solution for real-world applications.

Capsule Networks (CapsNets) and Long Short-Term Memory (LSTM).

"CAPSULE NETWORKS AND LONG SHORT-TERM MEMORY USAGE IN DETECTION OF IMAGE SPAM", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.6, Issue 4, page no.687-692, April-2019, Available :http://www.jetir.org/papers/JETIR1904W86.pdf

"CAPSULE NETWORKS AND LONG SHORT-TERM MEMORY USAGE IN DETECTION OF IMAGE SPAM", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.6, Issue 4, page no. pp687-692, April-2019, Available at : http://www.jetir.org/papers/JETIR1904W86.pdf