Publications

Conference Papers

PriBan: A Benchmark Dataset and Modeling Framework for Privacy Preservation in Bengali Texts

Anowarul Faruk Shishir, Bishaw Kirti Chakma, Indrajit Gupta
28th International Conference on Computer and Information Technology (ICCIT 2025)
IEEE

Ongoing Research Projects

Hierarchical Hybrid Workload Orchestration for Edge–Fog–Cloud CDN

In Progress

Designing a multi-timescale forecasting and provisioning framework for CDN traffic using edge burst detection, fog-level mixture-of-experts forecasting, and cloud-assisted continual learning. Aims to reduce SLA violations, latency, and operational cost under non-stationary workloads.

Fast–Accurate Online Learning for Edge AI

In Progress

Developing a lightweight online learning algorithm that bridges the speed–accuracy gap between OS-ELM and OR-ELM. Enables stable edge inference under drift and bursty workloads for Edge AI applications.

Undergraduate Thesis

Sentiment Analysis from a Bengali Dataset Using Hybrid Deep Learning and Contextualized Word Embedding Techniques

2024

Utilized contextualized Bangla BERT word embedding techniques to replace traditional methods (Word2Vec, GloVe, FastText) combined with a CNN-BiLSTM model, achieving 96.84% accuracy. Compared performance with 8 machine learning and 2 deep learning models, and introduced a feature for expressing sentiment with relative sentiment scores.