Machine Learning Projects (Python, TensorFlow, scikit-learn, Keras)

May 2018 – Present

  • Implemented several machine learning projects on Kaggle using multiple machine learning models including LR, KNN, SVM, DT, RF, XGBoost, Adaboost, GBDT, NB, NN, CNN etc.
  • Please see GitHub for more details

Reproduce MTCNN (Python, TensorFlow)

Jun 2018 – Dec 2018

  • MTCNN is a Multi-task cascaded convolutional network for face detection and alignment
  • Built the model structure using TensorFlow and implemented the process of convert the ground truth

    landmarks to tfrecord file with data enhancement

  • Implemented the process of calculating the loss of classifications, bounding boxes, and landmarks,

    finally achieved 93% accuracy

Build Face Recognition website (Python, TensorFlow)

Dec 2018 – Jan 2019

  • A face recognition website that can compare whether the person in the two pictures is the same person using python Flask as the user interface.
  • Use pre-trained MTCNN to detect the face position in pictures and use pre-trained FaceNet to calculate the distance between two faces.

Optimize DIVA Model (Python, TensorFlow)

Aug 2017 – Dec 2017

  • Optimized the hyperparameters selection in DIVA model (divergent auto-encoder model of category learning, an artificial neural network that uses reconstructive learning to solve N-way classification tasks) by using hyperopt approach. Shorten the search time at least doubled.