Ngoc Duy Tran

Personalise Theme

Introduction

I have a great passion in Machine Learning in general, Deep Learning in specific. So far, my research domains in Deep Learning include Computer Vision (Vision Transformer), Natural Langugage Processing (BERT, RoBerTa), and Time-Series Forecasting (N-BEATS). In addition, I am also interested in researching and applying machine learning and statistical models in Bioscience.

MORE ABOUT ME

I'm passionate about automation and innovative ideas. During my childhood, I was a huge fan of Minecraft; however, unlike other players who would manually collect and grow crops, I had a lazier approach: automating processes using a “redstone”, or electrical wire. From that moment, I have a dream of automating some parts of my everyday life like I had in my online world, which leads to my passion for technology. And BINGO! I’m now pursuing Data Science at Unimelb!!!

I am someone who finds joy in applying machine learning to create a positive impact on society. With the guidance of Dr. Gulrez Chahal and Professor Mirana Ramialison from the Murdoch Research Children's Institute, I am making progress in early detection of diseases for infants by applying machine learning methods on genes.

I also gain hands-on experience in building object-oriented and agile software systems in Java following design patterns to ensure future sustainability.

TOP EXPERTISE

Feel free to reach out if you want to connect! Download Resume

  • Python
  • C
  • R
  • Java
  • SQL
  • Matlab
  • Leadership
  • Problem-solving
  • Research skills
  • Team-work
  • Adaptability
  • Critical thinking

Computer vision

German Traffic Sign Recognition Benchmark

The German Traffic Sign Benchmark is a multi-class, single-image classification challenge, aiming at classifying a street sign given its image.

Read More
Facial Beauty Evaluation with Vision Transformer (accepted paper)

Facial Beauty Prediction technology also has applications in other areas, such as advertising and social media, where it can be used to optimize marketing strategies and help individuals enhance their online presence. This study introduces an effective approach to evaluate human face beauty using a transformer-based architecture.

Read More
Classification of 400 birds species

This project aims at classifying 400 bird species from a Kaggle dataset

Read More
American Sign Language classification

The data set is a collection of images of alphabets from the American Sign Language, separated in 29 folders which represent the various classes. This article aims at classifying these categories with an accuracy of 99%

Read More

Seq2Seq Models (NLP + Time Serires Modelling)

Book rating prediction

This project aims at evaluating book rating classification through 2 approaches: traditional machine learning methods via LGBM- Classifier[3], and deep learning method via transfer learning with BERT (Bidirectional Encoder Representations from Transformers)

Read More
Time-Series Forecasting of PM2.5 with N-BEATS

This paper focuses on a multivariate time series forecasting algorithm for air quality performed on Beijing Air Quality Dataset by using machine learning and deep learning models. More importantly, this paper also implements multivariate N-BEATS architecture - a state-of-the-art model, which is a pure neural network that beats Sequence to Sequence(seq2seq) models such as ES-RNN in its original paper

Read More
Machine Translation

This personal project applies transfer learning along with transformer-based architectures to perform one of the usual tasks: Machine Translation

Read More

Data analysis

Street Degraves Recycling Center

This article suggests the best pathway to help Melbourne CBD improve its waste management and sustainability by exploring Street Degraves Recycling Center's waste capacities and operations.

Read More
Titanic dataset

This article completes the analysis of what sorts of passengers were likely to survive in the sinking of RMS Titanic. In particular, tools of machine learning are applied to predict which passengers survived the tragedy.

Read More
London Bike Sharing Service

This guide will show you how to use Multivariate (many features) Time Series data by building a Bidirectional LSTM Neural Network to predict future demand

Read More

Object Oriented Programming (PacMan)

PacMan

This repo implements the normal version of PacMan and the extended version of it: Multiverse following GRASP rules and satisfying a set of pre-defined rules.

Read More
PacMan torusverse extension

This repo is an add-on to to the previous project, in which it adapts other design patterns such as Composite, Strategy. It also adopts better extensionability and implements a range of algorithms for PacMan to auto-run without user's intervention.

Read More

Contact with me