Projects

Gamification of Spaced Repetition Software

• Built a gamified spaced repetition software “NekoAnki”
• Performed a user-study to evaluate whether gamification improves user learning experience and entertainment value
• Implemented a framework for a mobile interface based reward system for gamification purposes

Continuous Word Recognition

• Performed an HMM-based speech recognition task on the TIMIT speech dataset using the Kaldi toolkit
• Wrote shell scripts to investigate the influences of acoustic features and dynamic features on word error rate

Image Captioning on MSCOCO dataset

• Used the TensorFlow framework to perform an image captioning task on the MSCOCO dataset
• Constructed an encoder-decoder framework with a CNN encoder and an RNN decoder with GRU and LSTM
• Rewarded for third best project in Machine Learning Practical by IBM

Image Classification on the MNIST and EMNIST datasets

• Classified images of handwritten digits and letters from the MNIST and EMNIST datasets using deep neural networks
• Implemented models with different activation functions, weight initialisation strategies, and learning rules

Speaker Independent Digit Recognition

• Recorded voiced English digits and parameterised the collected waveform files as MFCCs
• Constructed a speaker-dependent, and -independent speech recogniser using the Hidden Markov Model Toolkit (HTK)

Facial Expression Prediction - Kaggle Competition

• Used the Toronto Faces dataset to classify faces based on several facial expressions