2021/01/11
Contribution from kuancalvin2016@gmail.com
Computer Vision Research - Semantic Segmentation in Land Classification (2021) Code Link
Computer Vision Research - Object Detection in House Detection (2021) Code Link
Computer Vision Research - Semantic Segmentation in House Detection (2021) Code Link
(Due to the confidential issues, only some codes have been uploaded)
• Web Crawler for Taiwan Stock including financial report & individual stock report
• Research for ML Stock Price Prediction - LSTM, Seq2Seq, Attention, WGAN-GP, WGAN, GAN
Code Link - GAN, Crawler, A3C Trading Robot
(Due to intellectual property, the code cannot be uploaded)
• Visualisation through python packages, such as dimension reduction - tSNE
• Text Mining
• Paper research of semi-supervised learning, BERT, Variational Autoencoder Classifier, Data Augmentation
• Implementation of new models from latest papers, including Spacy NER model, BERT Fine-tune model, word- based entity embedding BERT model, sentence-based BERT model, BERT LM unsupervised learning and so on
• Compared with different NN structures, incl. Bi-LSTM, CRF – F1 score 83.2% and 96%
• Utilised different embedding function, incl. Glove and Tensorflow Elmo
Code Link
• Model 1 - Glove + Global & Soft Attention + Bi-LSTM, 31d multi-labels - RMSE 0.83
• Model 2 - XLNet Fine Tune + Downstream, 31d multi-labels regression - RMSE 0.80
• Model 3 - XLNet Fine Tune + Downstream, 1d single label regression - RMSE 0.5
(Due to the confidential code from Oleeo, these files are examples for the projects)
Code Link
• Compared with different structures, incl. Bi-LSTM + Global Soft Attention and SVM – Acc 0.771
Code Link
• Utilised Bi-LSTM to create language speech recognition system transforming audio data to text
• Audio → MFCC → NN → Target: CTC Loss (Beam Search, Edit Distance)
Code Link
• Achieved top 6% and accuracy of 0.11 for the RMSE
• DNN, Linear Regression & Stacking Model (Lasso, Elastic Net, SVR, Kernel Ridge, Bayesian Ridge, Ridge)
Code Link
• Two CNN structures, incl. Fastai DenseNet 201 & NASNet + global max/average pooling – Acc 95.9%
Code Link
• Model 1: Utilised NN to create a price prediction system
• Model 2: Utilised Latent Dirichlet Allocation to create a topic system
• Model 3: Utilised Google Cloud Vision API to create an image recognition system
• Workflow: Created ML models (Python) → Imported 3 models to API → Created a mobile application
Code Link
• Market Demand: Theft offences occurred at a rate of 45.8 per 1,000 population in 2017 in London
• Workflow: Dimension Reduction → ARIMA & LSTM → Validation → Output Prediction
• Function: Once users walks on a road at a specific time, they will be reminded if the time is dangerous
• Achieved top 2% and accuracy of 83.7% for the prediction of the data
• Utilised 2 methods to predict labels respectively, incl. DNN & Random Forest
Code Link