Hi! I'm Phi. I am a PhD candidate in Computer Science in Nanyang Technological University, Singapore (NTU). I work in Neural Natural Language Processing under Prof. Shafiq Joty. I focus in various aspects of Neural Machine Translation methods, especially unsupervised and semi-supervised machine translation.
This summer 2021, I begin my Research Internship at Facebook AI Research (FAIR) under Dr. Hongyu Gong. My work there involves Machine and Speech Understanding and Translation.
Find my CV here.
My research involves various topics of Deep Learning in Neural Machine Translation (NLP). I try to innovate and invent novel models and architectures, such as the Transformer, to improve the performance of data-driven machine translation systems. I also explore numerous techniques to enhance unsupervised and semi-supervised machine translation systems, such as using diversification methods. My work can meaningfully contribute to the development of commercial translation systems like Google Translate.
Plus, I work on side research projects regarding semantic parsing, summarization, etc.
- Multi-lingual NLP (Machine Translation, Cross-lingual tasks)
- Machine Learning
- Deep Learning
- Reinforcement Learning
May 2021 – Sep 2021
United States (Remote)
Research on Machine Translation, Automatic Speech Recognition (ASR) and Speech Translation.
May 2019 – Aug 2019
NLP Research Intern
Research on different aspects of linguistic structures languages on the performances of neural architectures on various natural language processing tasks. Propose new state-of-the-art methods and write papers submitted to machine learning and NLP conferences.
Mar 2018 – May 2019
Research on different limitations and improvements on Neural Machine Translation, such as document-level machine translation, discourse phenomena, phrase-based, parsing-tree-based and unsupervised neural machine translation. Write papers submitted to various machine learning and NLP conferences, e.g: ICLR, ACL, EMNLP.
May 2017 – Jul 2017 & Jan 2018 – Jul 2018
Software Engineer Intern
Developed a novel lightweight character-level convolutional neural network to perform scripted text classification tasks at up to 98 accuracy while consuming 1000 times less resources and achieving 10 times faster training time than standard deep models. Developed production-level code to deploy the models.
May 2016 – Jul 2016
Software Engineer Intern
Researched and cooperated to develop a new machine learning algorithm based on Support Vector Machine to classify electrical signals, achieving 94% of experimental accuracy. Assisted to design a Raspberry Pi robot for collecting sensor signals and communicating with server to manipulate a real car's system. Real-time accuracy reached 83.8%.
Doctor of Philosophy in Computer Science & Artificial Intelligence
Research on neural approaches for supervised/unsupervised/semi-supervised machine translation systems.
Bachelor in Electrical & Electronics Engineering
Worked in various machine learning projects, ranging from classical ML like SVM to deep learning methods for computer vision and natural language processing.