Koay Xian Cong

"At the beginning of the course, I only know the basics of Java programming and I did not think I will be able to catch up. However, with the constant help from tutors, friends and replays of the classes, I manage to cope with the pace and passed the exam!"
Program Taken:
Applied Deep Learning Bootcamp Graduate (Nov'20)
Example of my work:
Click here to view my work
View program details

With the rapid innovation happening in the electronics, science and communication network, our lives have improved significantly especially when these technologies converge. Undoubtedly, I aspire to contribute to this megatrend of engineering advancement one day. 

To support my aspiration, I applied to Forward School to join their Applied Deep Learning Bootcamp. I learned how to build Artificial Intelligence Neural Networks using Java language and applying it to Computer Vision that gained high-level understanding from images. The tools I used throughout the whole course is DL4j libraries and IntelliJ IDE.

During the course, I built an acne classifier AI model together with Han Sheng and LiShan that analyzes images to classify various users’ acne types. As we all are aware, acne is an issue faced by most teenagers and even adults. However, there is no proper way of classifying acne types to get treatments based on the different levels of seriousness and categories. Thus, we decided to train an acne classifier using the dataset we found from Kaggle. It was a time- consuming and challenging project to train, test and evaluate the model.

My first DL4J project
"At the beginning of the course, I only know the basics of Java programming and I did not think I will be able to catch up. However, with the constant help from tutors, friends and replays of the classes, I manage to cope with the pace and passed the exam!"

We spent 3 weeks building our own model and found that the accuracy is not high enough. Thus, we moved on and tried different architecture such as Resnet-50 and VGG-16. After around 50 trials, we decided to use VGG-16 as our base model. I learnt many different algorithms and architecture that can increase output accuracy. Finally, we fine-tune the model and achieve 74% accuracy.

At the beginning of the course, I only know the basics of Java programming and I did not think I will be able to catch up. However, with the constant help from tutors, friends and replays of the classes, I manage to cope with the pace and pass the Applied Deep Learning Examination in Computer Vision and get offered by Skymind to work as a Junior Deep Learning Engineer.

I am looking forward to contributing the knowledge I learnt from the Bootcamp in solving real-world problems in the near future.