Young D. Kwon

Subspace Clustering Algorithm of SubKmeans

Studied the original paper by implementing the algorithm from scratch and further applied other subspace clustering algorithms on various datasets
Course project at HKUST
[Code | Report]

Mastering Scikit-learn

Explored the performances of various machine learning algorithms using Scikit-learn software package on different datasets. Compared 4 ML models such as Logistic Regression, Linear SVM, SVM with RBF kernel, and Neural Networks.
Course project at HKUST
[Code]

Beer4You

Implemented an Android app which presents information on beers to users when they capture beers' logo using the image recognition. The app further recommends beers to users based on their previous search histories
Excellence Award (1st Runner-up, awarded US$1,000)
Competition, Conn App 2015
[Code | Slides]

Wearable of the future: ABOUT AR

Collaborated with a team of seven members from the various background such as the fashion design, product design, and computer science to develop a future product design in regard to augmented reality
International Workshop at Samsung Art and Design Institute 2015
[Slides]

Deep Learning Framework for NLP from scratch

Implemented C++ based framework for deep learning research from scratch. Implemented function codes for derivatives, neuron units (e.g., sigmoid, tanh, relu), loss functions (e.g., negative log likelihood), matrix operators, parameter initialization (e.g., Xavier, Randomized), dealing with datasets, forward and backward propagations. Experimented my framework with well-known NLP task, Named Entity Recognition.
Course project at SKKU
[Code | Slides]

Server Queue Scheduling Simulation

Analyzed conditions which cause bottle neck situation in a simulated environment. Improved the efficiency of server queue scheduling with limited resources
Course project at SKKU
[Code | Slides]

Statistical Blackjack AI

Implemented an blackjack game and AI by statistically modeling Hit and Stand probability. The implementation is based on C++
Course project at SKKU
[Code | Slides]

Sliding Window Face Detection

Distinguish actual face objects from non-face objects using Histogram of Oriented Gradients (HOG) and Support Vector Machine (SVM). Implementation is based on Matlab
Course project at SKKU
[Code | Report]

Parallel Programming on 3 Dimensional Map Array Using Pthread API

Significantly improved the speed of the highly optimized sequential program (based on B+ tree) using Pthread API and Parallel programming. Minimized overhead by reducing Mutex Lock (less shared memory by all processes), memory access (1D long array instead of 3D map), and using buffering (resolving dependency problem)
Course project at SKKU
[Code | Slides]

MPhil Projects