chow siew yeng

Chow Siew Yeng (3)
Q: Share something about yourself.

Hello, I am Siew Yeng. I’m currently doing a Masters degree in Linguistics at Nanyang Technological University (NTU). I research on Singlish grammar, and I write certain computational rules on it. Grammar is a very interesting subject - It’s not about telling people how to speak, but rather looking at the way they use language and figuring out the rules that govern it.

It is very common to hear people say that Singlish is kind of “anyhow” but you definitely can’t just randomly string some English words together and pepper in Hokkien or Malay and call it Singlish because there are certain constraints. That's why, I wanted to formalise some of those rules for this language variety I grew up speaking. I started my internship at A*STAR's 
Institute of High Performance Computing (IHPC) around mid-January and it’ll end in March 2021.

Q: Tell us about what you do at IHPC?

I was involved in two projects - K-EMERGE and one on Multimodal facial landmark detection. 

In project K-EMERGE, I offered my opinion as a linguist to the commonsense Natural Language Processing (NLP) portion of the project and researched on discourse representation and benchmarks in the NLP community.

In the other project, I built a Multilayer Perceptron (MLP) model and a Convolutional Neural Network (CNN) model to detect emotion based on certain landmarks found on the face.

The model that I built using CNN has multiple layers. I used Google Mediapipe to extract the landmarks from the images of faces obtained from Kaggle and then fed them into the model, together with the labels that corresponded to the emotions. (P.S. so I don’t actually judge the emotions what each face shows!) I spent most of the time uploading the data and finding the best set of parameters for the model to train on to increase its accuracy. 

I was pretty lost at the initial stage as this was completely new to me and I spent a lot of time on troubleshooting. I even asked a friend to send me his CNN assignment from the previous semester even though it used a different library. But it felt really fulfilling and exciting when I successfully trained the model for the first time.

Q: What/Who is your inspiration in life?

I’m inspired by the people who seem so driven in life and enjoy it very much.

One of them was someone I met two years ago as a volunteer on a farm in Ireland. He is a very interesting and happy person but what stood out to me most was his goal at that moment. He lived in a place where he could see, just behind his house, the adverse effects of climate change. The forest was dying because of the drought. He was a strong advocate for sustainable living and was not satisfied with how the education system there did not teach that. Together with some friends, he aims to set up a school to teach children and teenagers about the concept of sustainability and how to grow their own food. And volunteering at the age of almost 40 to learn about farming was one of the steps he took towards that.

It is amazing to see people change their path in life based on such a good cause. I admire the determination and the bravery it takes to do something like that and I’m inspired to be part of the change that I want to see.

Chow Siew Yeng (2)

Q: Describe a typical work week.

When I first started the internship, I tried to work on alternate days for each of the two projects. After running some codes for the CNN, I realised there’s a lot of waiting thus the first thing I did when I woke up every day was to run certain code or upload images. I would then read up on papers related to the commonsense project while waiting for the code to be completed. As working from home is still the default mode of work, I would occasionally return to the IHPC office for meetings and usually end my workday between 4.45pm to 7.30pm, depending if I have to prepare dinner or any activities that I need to attend. 

Q: How has this internship benefits you?

I’ve definitely learnt a lot during this internship from gaining skills like how to build a neural network model and understand how a project with different parts can come together. I think I know better now what to expect if I join a company similar to IHPC in the future. I have also met many nice, interesting, and patient people from IHPC, both while working and while playing soccer.