UNIVERSITI TUNKU ABDUL RAHMAN

Postgraduate Final Year Project Titles

Updated on 21/2/2025


   No   Supervisor   Co-Supervisor    Title    Skills/Tools    Industry Linked?    Status    Brief Description
  1  Png Wen Hao  Not Available Mini LLMs Orchestration for Efficient Multi-Expert Agentic Workflow
  1. Python programming for Data Science
  2. May involve extensive uses of platforms and libraries like LangChain, LlamaIndex, HuggingFace, and PyTorch
 No       
project details ⊳

Existing LLM-centric workflow designs, which heavily rely on LLMs for various reasoning and actions, are often onerous and inefficient. Empirical evidence shows that such workflows typically require pageful prompts with detailed instructions and few-shot examples to enforce stepwise reasoning or course of actions in completing the designated task. This project shall focus on the mixture-of-expert (MoE)-like techniques to facilitate the long chain of actions using multiple tiny LLMs.
2 Khor Siak Wang Not Available Object Detection and Recognition
  1. Mathlab

  2. C++

  3. Java
 Preferably yes       
project details ⊳

To develop a system that could help detect and recognize the presence of an object. It's particularly useful in detecting some unusual activities, e.g. abandoned objects, intruders, vandalism, snatch thieves etc.
3 Kiran Adnan Not Available Enhancing Cybersecurity through Advanced Threat Detection and Response Techniques
  1. NVivo
 Yes       
project details ⊳

This project aims to identify and assess advanced threat detection and response techniques. Also, it aims to explore the integration of new techniques with existing cybersecurity frameworks. And, to identify the challenges and limitations associated with the implementation of advanced threat detection and response techniques.
4 Aun Yichiet Png Wen Hao Carbon footprinting on Ethereum Blockchain
  1. Ethereum
 To be extendeded into...       
project details ⊳

To profile consumer products carbon footprint based on consumer consumption behaviors.
5 Ramesh Kumar Ayyasamy Not Available Developing a sentiment analysis model using generative pretrained transformer for education domain
  1. Python programming
  2. Basic knowledge in deep learning
  3. Text mining techniques
No       
project details ⊳

This project aims to develop a sentiment analysis model using the generative AI techniques to evaluate ChatGPT-related tweets in education contexts.The categorization and analysis of thematic content within tweets provide valuable context for understanding the specific areas of education where ChatGPT is generating interest and discussion.
6 Lim Seng Poh Not Available Performance Analysis of Differential Evolution on Benchmark Functions
  1. C++ programming
 No       
project details ⊳

Differential Evolution (DE) is one of the popular evolutionary algorithms. Some of the previous works did not consider identifying suitable generation schemes and control parameters for producing trial vectors. Consequently, a longer duration is needed to produce good results. With the increasing application of DE in solving optimisation problems, the performance of DE becomes an interesting topic to explore. Therefore, it is essential to study the performance of different DE models before applying them in specific case studies.
7 Muhammad Syaiful Amri bin Suhaimi Not Available Camera-based eye gaze and head movement tracking for online learning
  1. Python

  2. Matlab

  3. any relaven computer vision programming
 No       
project details ⊳

The main objectives is to track the eye pupil or gaze using computer vision/image processing during an online class or a meeting. There are several challenges for the project:
  1. The accuracy of gaze tracking using computer vision techniques
  2. The effect of head/body movement to the tracking accuracy.
  8  Goh Chuan Meng Not Available Development of Deep Learning algorithms for cell death prediction in in-Vivo imaging
  1. Python
  2. Matlab, or any relevant computer vision programming
 Preferably yes       
project details ⊳

This research aims to create advanced computational tools that can analyze in-vivo medical images to predict and identify instances of cell death. Cell death is a crucial aspect in various biomedical and clinical research areas, as it can provide insights into disease progression, treatment effectiveness, and tissue health. This research involves the design, training, and validation of deep learning models, such as convolutional neural networks (CNNs) or recurrent neural networks (RNNs), to process complex in-vivo imaging data. The ultimate goal of this research is to develop accurate and robust algorithms that can not only detect cell death events but also predict them in real-time or over specific timeframes.
  9  Goh Hock Guan Not Available IoT Agriculture Monitoring using DeepSeek
  1. Python
 No       
project details ⊳

Using DeepSeek AI on the Raspberry Pi to monitor the growth and prediction for Agriculture.
https://youtu.be/o1sN1lB76EA?si=vZGFSpLntpNj8OoA
  10  Adeb Ali Not Available Deep Learning-Based CPU Scaling and Renewable Energy Harvesting for Sustainable 6G Edge IoT
  1. Python
  2. Matlab programming
 -       
project details ⊳

Deep learning enables AI-driven CPU scaling and renewable energy harvesting for sustainable 6G edge computing in IoT networks. By optimizing CPU frequency based on workload demands and utilizing solar, RF, and ambient energy, IoT devices achieve energy efficiency, extended lifespan, and reduced carbon footprint, supporting next-generation green edge computing.
  11  Chai Meei Tyng Not Available Multimodal Emotion Recognition for Depression Detection
  1. Python or Matlab
 No       
project details ⊳

This project will integrate multiple data sources such as facial expressions, speech tone, and text analysis to improve the accuracy of depression detection. A machine/deep learning model will combine these features to provide a more comprehensive emotional assessment.
Designed by : 2024 Syed Muhammad DIS FICT UTAR