Ashraf Ul Alam

Download PDF

Education

Rajshahi University of Engineering & Technology (RUET), Rajshahi, Bangladesh
Bachelor of Science in Computer Science and Engineering
2019 - 2024 | CGPA: 3.44 out of 4.00

Relevant Coursework: Neural Networks & Fuzzy Systems, Data Mining, Artificial Intelligence, Digital Image Processing, Database Systems, Parallel and Distributed Processing, Digital Signal Processing, Data Structure (C), Object Oriented Programming (C++, Java), Computer Algorithms, Applied Statistics & Queuing Theory (Python) <!– —

Standardized Test Scores

IELTS: Overall 7.0 (Listening: 8.0, Reading: 6.5, Writing: 7.0 & Speaking: 6.5) –> —

Experiences

Young Learners’ Research Lab

Research Assistant (March 2023 - May 2024)
Lab Head and Supervisor: Md. Azmain Yakin Srizon

Key responsibilities included designing optimized CNNs for medical image analysis, constructing OCR datasets, and developing algorithms for extracting handwritten Bengali text. Additionally, published two conference papers in Taylor & Francis and IEEE.

Publications

Journals

1. Ashraf Ul Alam, S. M. Hasan, S. P. Islam, and M. P. Uddin, “PRODIA: A Probability Distribution Alignment Framework for Unsupervised Domain Adaptation in Medical Image Segmentation,” Computers in Biology and Medicine, Under Review.

Conferences

1. Ashraf Ul Alam, S. P. Islam, S. M. Hasan, A. Y. Srizon, M. F. Faruk, M. Al Mamun, and M. R. Hossain, “Optic Disc and Cup Segmentation via Enhanced U-Net with Residual and Attention Mechanisms,” 2024 6th International Conference on Electrical Engineering and Information & Communication Technology (ICEEICT), IEEE, 2024, pp. 329–334, doi: 10.1109/ICEEICT62016.2024.10534436. [Nominated for Best Poster Award at ICEEICT 2024]
2. S. P. Islam and Ashraf Ul Alam, “Advancing Ophthalmology through Transfer Learning and Channel-Wise Attention for Retinal Disease Classification,” 2024 6th International Conference on Electrical Engineering and Information & Communication Technology (ICEEICT), IEEE, 2024, pp. 313–318, doi: 10.1109/ICEEICT62016.2024.10534342.

Book Chapters

1. S. Das, Ashraf Ul Alam, S. P. Islam, A. Y. Srizon, “BanglaOngko: A New Dataset for Accurate Bengali Mathematical Expression Detection Utilizing YOLOv8 Architecture,” Data Driven Applications for Emerging Technologies, 2nd International Conference on Big Data, IoT and Machine Learning (BIM 2023) Taylor and Francis, pp. 43-57, Proofreading Stage.

Undergraduate Thesis

KD-UDA: Knowledge Distillation-based Unsupervised Domain Adaptation for Improved Medical Image Segmentation [Project]

Tensorflow, Keras, CNN, Transfer Learning, U-Net

• Developed the KD-UDA framework, using Knowledge Distillation to enhance segmentation model performance on diverse medical imaging datasets without labeled data from new domains, significantly improving performance for both 2D retinal fundus images and 3D MRI data (BraTS2021).

Projects

Cyber Threat Report Summarization Using FLAN-T5 with LoRA Adaptation [Project]

Pytorch, FLAN-T5, LoRA, Transformers, Text Summarization

• Implemented a cyber threat report summarization model using FLAN-T5 for text generation. Applied LoRA to enhance fine-tuning efficiency, enabling better extraction of critical threat information while preserving important details in the summaries.

Parameter Efficient Fine-tuning of DistilBERT with LoRA for Phishing URL Detection [Project]

DistilBERT, LoRA, Transformers, Text Classification

• Fine-tuned a DistilBERT model for phishing URL detection, incorporating LoRA adaptors to significantly enhance performance, while optimizing the model for efficient use in resource-constrained environments.

Cycle Thief Detection from Realtime Footage using YOLOv5 and DeepSORT [Project]

OpenCV, YOLOv5, DeepSORT, KD-Tree, Face_Matcher

• Developed a real-time cycle thief detection system utilizing YOLOv5 for object detection, DeepSORT for tracking, KD-Tree algorithm for efficient nearest neighbor search, and Face_Matcher for facial recognition from live CCTV footage.

NeuroSeg3D: 3D Attention U-Net for Accurate Brain Tumor Segmentation (BraTS 2021) [Project]

3D U-Net, Residual Blocks, Spatial Attention

• Developed the NeuroSeg3D architecture, enhanced with residual blocks and spatial attention modules, to achieve accurate brain tumor segmentation, and demonstrated the model’s performance on the BraTS 2021 dataset.

Chronic Kidney Disease Prediction using Machine Learning [Project]

Python, Flask API, HTML, CSS

• Performed comprehensive exploratory data analysis and feature engineering to enhance the accuracy of a CKD prediction model. Deployed the model using Flask API and designed a user-friendly web interface with HTML and CSS for CKD risk assessment.

Optimizing Feature Representation of Deep Neural Networks for Enhanced Deepfake Detection [Project]

VGG16, Channel Attention, ResNet50, Deepfake

• Focused on detecting deepfake images using the 140k Real and Fake Faces dataset. Employed VGG16 as the primary feature extractor, then introduced a channel attention mechanism to prioritize relevant feature vectors and finally achieved a high classification accuracy of 99.80%. Further, conducted an ablation study with ResNet50.

Cardiotocogram Data Analysis for Fetal Health Classification Using Machine Learning [Project] [Slide]

Random Forest, K-Nearest Neighbors, Gradient Boosting, SMOTE

• Aimed to classify fetal health status from Cardiotocogram (CTG) data using machine learning models. Utilized a dataset of 2,126 records with 21 features. Applied models like Random Forest, K-Nearest Neighbors, and Gradient Boosting, achieving the highest accuracy of 98.47%. Employed feature selection, data standardization, and SMOTE to enhance performance.

Maternal and Child Health Care [Project]

HTML, CSS, PHP, MySQL, Android Studio, Java, XML, Firebase Database

• Developed a responsive website for Maternal and Child Health Care featuring due date calculations, immunization schedules, personalized notifications, and a query posting feature. Created a mobile app using Android Studio and Firebase with the same features.

Implementation and Analysis of Neural Networks for Liver Disease Diagnosis [Project]

KNN, Single Layer Perceptron, MLP, Kohonen Self-Organizing Map, Hopfield Neural Networks

• Implemented and analyzed various neural networks and machine learning algorithms from scratch to gain a deeper understanding of how these models function. The project focuses on classifying liver disease using the Indian Liver Patient Dataset and explores a variety of learning techniques.

RSMS: Retail Store Management System [Project]

Android Studio, Java, XML, Firebase Database

• Implemented a mobile application to streamline retail store operations, including inventory management and sales tracking.

e-Doctor’s Appointment [Project]

Android Studio, Java, XML, Firebase Database

• Developed a mobile application for managing and scheduling medical appointments, enhancing user convenience and appointment organization.

Technical Skills and Interests

• Research Areas: Computer Vision, Domain Adaptation, Object Detection, NLP, LLM, Transfer & Conventional Learning
• Programming: Python, C, C++, Java, PHP
• Frameworks: TensorFlow, Scikit-Learn, Keras, OpenCV, PyTorch, Bootstrap
• Web & Databases: HTML, CSS, PHP, MySQL
• Technologies: Flask, Android Studio, LaTeX, Git

References

S. M. Mahedy Hasan (Undergrad Thesis Supervisor)
Assistant Professor
Dept. of Computer Science & Engineering
Rajshahi University of Engineering & Technology
Mobile: +880-1870100318
Email: mahedy@cse.ruet.ac.bd

Md. Azmain Yakin Srizon (Project Supervisor)
Assistant Professor
Dept. of Computer Science & Engineering
Rajshahi University of Engineering & Technology
Mobile: +880-1790187189
Email: azmainsrizon@gmail.com

Contacts

Phone: +880 1868-406894
Email: ashrafamit9227@gmail.com
LinkedIn: linkedin.com/in/ashraf-ul-alam-amit
Address: 25/2, Nabakalash, Matlabganj, Matlab South, Chandpur-3640, Bangladesh