I am Ananta Raha, a CSE graduate with a strong research interest spanning multidisciplinary applications of Computer Vision, Deep Neural Networks, Medical Imaging, Explainable AI, and Multimodal Learning. Particularly, I am motivated by developing trustworthy AI systems that ensure explainability, transparency, and efficiency.

Following my undergraduate thesis on computationally-efficient brain tumor segmentation (from MRI), I have researched lightweight CNN architectures in medical diagnosis and precision agriculture. During my master's studies, I worked on independent and team projects in diverse areas like automatic malaria diagnosis, edge-ML for sericulture, neural network privacy, RNA-seq read disambiguation, and human-computer interaction. These experiences have reinforced my skills and perspective on practical ML applications.

Currently, my master's thesis focuses on multimodal fusion (Text + Vision) learning to improve radiological segmentation tasks. Additionally, I am actively seeking Ph.D. opportunities where I can contribute to quality research in on-device, trustworthy neural networks for applications in healthcare and precision agriculture.

• IELTS: 7.0 (Listening 7.0, Reading 7.0, Writing 6.5, Speaking 6.5)

• Raha, A. Lightweight COVID-19 Detection from Chest CT-Scans Using Attention-Based CNN. SN Computer Science, Vol. 6, 853 (2025). DOI: 10.1007/s42979-025-04403-5 [Read Online]

---

• Raha, A. and Q. S. Tasnim. Efficient Mulberry Leaf Disease Detection in Bangladesh: A Lightweight Approach for Real-Time Applications. 2025 International Conference on Electrical, Computer and Communication Engineering (ECCE), Chittagong, Bangladesh, 2025, pp. 1-6, DOI: 10.1109/ECCE64574.2025.11013948 [Download Preprint]

---

• Raha, A., Parvin, F., Jannat, T. (2024). Brain Tumor Segmentation with Efficient and Low-Complex Architecture Using RCNN and Modified U-Net. In: Arefin, M.S., Kaiser, M.S., Bhuiyan, T., Dey, N., Mahmud, M. (eds) Proceedings of the 2nd International Conference on Big Data, IoT and Machine Learning. BIM 2023. Lecture Notes in Networks and Systems, vol 867. Springer, Singapore. DOI: 10.1007/978-981-99-8937-9_22 [Download Preprint]

• Hossain, M., Sultana, A., Datta, S., Raha, A., Khan, S., Ahmed, R., Sharmin, S. (2025) Analyzing Gamification in E-learning Tools: HCI-Driven Insights from University Students in Bangladesh

---

• Raha, A., Ahmed, J., Hossain, M.S., Majumdar, S. (2025) Defending Model Inversion Attack Using an Improved Filter-Based Approach

2025 Postgraduate Collaborative [View Source]

Keywords: Sequencing, Gene Expression, Illumina, Computational Biology, Short Reads, MLP

NGS technologies produce short reads that are aligned to a reference genome for downstream analysis. In paired-end mode, some reads map to multiple loci, introducing ambiguity. This project minimizes that issue with an MLP-based filter that predicts and discards incorrect alignments of multi-mapping reads reported by the aligner. The approach derives context-aware features—including read coverage, RNA-seq, and k-mers. Evaluated on Bowtie2 alignments for two reference genomes (Escherichia coli and Bacillus subtilis), the trained classifier reduces the overall misalignment rate by substantial margins.

2025 Postgraduate Collaborative

Keywords: Human-Computer Interaction, Gamification, Learning, Social Interaction

Most online learning platforms in Bangladesh have reduced learner engagement due to various factors, including gamification, which is a major one. This project surveys 209 university students to filter out the learner’s perspective on gamified features and engagement. Afterwards, it inspects and analyzes the low-tier platforms to technically suggest the significance of each feature. These insights will help current and future learning platforms by systematically highlighting the evident factors in learner engagement.

2024 Postgraduate Collaborative [View Source] [Download Paper]

Keywords: Sericulture, Leaf Disease, Data Mining, Classification, Deep Learning

The project develops an efficient classification model to accurately identify mulberry leaf diseases, supporting sericulture in rural areas of Bangladesh. We utilized a regional dataset containing mulberry leaf images captured from Rajshahi, the principal sericulture region in Bangladesh. We performed explainability analysis and assessed the impact of preprocessing techniques on classification outcomes. While state-of-the-art models tend to be computationally expensive, our architecture, having only 145K parameters, remains surprisingly lightweight and outperforms the base MobileNetV2 model in our experiments.

2025 Postgraduate Collaborative

Keywords: Neural Network Security, Defense

The project develops an efficient defense mechanism against model inversion attacks that exploit the confidentiality of training data in neural networks. While existing techniques primarily require modifications to the trained models or even retraining, our robust approach does not alter existing architectures. Moreover, it works against various types of attacks, including the invincible label-only attacks.

2024 Postgraduate Collaborative [View Source]

Keywords: Blood Smear, Pathological Diagnosis, Classification, Deep Learning

This project presents a cascaded three-stage localization and classification architecture to effectively identify the lifecycle of malaria parasites in full-slide microscopic blood smear images. The novelty lies in its end-to-end automated workflow, while existing works require pathologist-aided cell extraction. In addition, it is less complex compared to expensive SOTA classification and object detection models.

2024 Independent Individual [Read Paper]

Keywords: Medical Imaging, Computer Vision, Deep Learning, Classification

A lightweight CNN architecture to identify COVID-19 in Chest CT scans. Despite having minimal trainable weights, it shows superior performance in the experiments.

2023 Undergraduate Individual [Download Paper]

Keywords: Tumor Segmentation, RCNN, U-Net, Less-complex

This project was a part of my undergraduate thesis. It develops an efficient technique for precisely segmenting tumor regions from brain MRIs. The two-stage approach consists of RCNN and a modified U-Net. The proposed system is low-complex with a small number of parameters and exhibits remarkable performance compared to other existing architectures.

2022 Undergraduate Collaborative

Keywords: Data Preprocessing, IoT, Embedded Systems

The project constructs a low-cost IoT-enabled system for continuous monitoring and nurturing of silkworms, which is crucial for producing silk in the sericulture industry. It also evaluates the applicability in potential regions in Bangladesh.

2019 – Present Hobby Android App [Google Play Store]

With my knowledge of OOP and Java, I developed this hobby project, which has become one of the most popular finance apps on the Google Play Store, with over 1 million active users worldwide. Over the years, I have consistently maintained it, which has allowed me to learn the fundamentals of software testing and lifecycle management.

2022 Hobby Android App [View Source]

Inspired by the idea of interpreting Morse signals at night, I developed this app to decode Morse code from blinking lights using the mobile camera. It seamlessly handles human timing errors to decode in real time, leveraging Android's CameraX library. It samples captured frames at a fixed rate, intelligently measures dot, dash, and space durations. Finally, it maps them to text using the Morse code table – all in real time.

2021 Undergraduate Individual Web Development [View Source]

Keywords: Project, Android App, Management

I developed this web-based software as part of the CSE 3100 undergraduate project. It simulates the workflow of a supermarket management tool, featuring dedicated terminals for cashier operations, inventory administration, and analytics. Built with Django as the backend and Bootstrap for the frontend UI, the system integrates core functionalities like billing, stock tracking, sales analysis, and authorization.

• Fellowship – BUET MSc Program (April 2024) in recognition of academic excellence.

---

• Education Board Scholarship (2017, 2015, 2012) for excellence in the HSC, SSC, and JSC Examinations, respectively.

---

• Specialization: Python for Everybody – University of Michigan (2020); [Verify Certificate]

---

• Regular Expressions in Python – Coursera (2022); [Verify Certificate]

---

• Participant at National Science Olympiad, Dhaka (2014) – Secured 34th position nation-wide.