Hnkty4kpokjxsvm1ihxh

Chenyu Tsai

A cross-domain practitioner, big fan of programming, linguistics and design thinking.

Main research area includes explainable NLP system, natural language understanding. Devoting to build more trustworthy AI system.

Additional research interests, human computer interaction, user experience, design thinking, writing.

NLP Engineer / Data Scientist / Machine Learning Engineer
[email protected]

Education

National Chengchi University, Master of Science(MS), Digital Content, 2018 ~ 2020

GPA: 4.2/4.3

  • Main research.
    • Explainable NLP system.
    • Natural language understanding.
    • Deep learning.
  • Additional research interests.
    • Human-computer interaction.
    • User Experience.
    • Design thinking.
  • Head TA of Artificial Intelligence and Design Thinking, coaching 50 students to develop 8 AI projects.
  • Won 1st place out of 70 teams in TRANS ACTION with 6 mates to develop an service that improves migrant worker environment in Taiwan. 

National Chengchi University, Creative Design Program, 2013 ~ 2014

  • Project Name: Byster
    • Unity Engine
    • Game Design

National Chengchi University, Bachelor of Arts(BA), Chinese Literature, 2011 ~ 2015

GPA: 4.0

  • Golden Melody, 2nd place.

Work Experience

Garden Villa , Special Assistant to General Manager, Jul 2015 ~ Aug 2018

  • Cooperating with General Manager in recruiting, training, payroll processing, performance.
  • Coordinating with General Manager in planning short and long term projects, budgets, expense controls, schedules and human resources.
  • Implementing quality and productivity objectives to achieve company goals.
  • Supervising staff and controlling merchandise.

Publication


  • Modeling Meta-Explainability of Natural Language Inference
    • Evaluating model explainability with human-involved test, to provide the Explainable Artificial Intelligence community cutting edge insights.
    • In Natural Language Inference, we identified attention mechanism do tend to attend the key information after trained with different but related tasks.
    • Retrieving explanation to the NLI task from attention weights, analyzing the meta-explainability of different models and methods.
    • Comparing the difference between how model and human make an explanation, evaluating human trust and preference to the model explanation.
    • Github Link 

Projects


Below are some of my project experiences. Most projects combine information technology and user research/experience to better meet users' needs.


Mygran

  • TRANS ACTION AWARD 2018 - 1st place out of 70 teams.
  • A competition combined cross-domain and user experience.
  • Proposing a service to solve problems derived from the  growing needs of migrant workers.
  • Taking part in the whole research progress and design the application.


NCCU Foods & Friends

  • Start from the inconvenience of the blind.
  • Find the true pain point - poor dining experience
  • Proposing a service to improve their dining experience. 
  • Taking continuity and universality into consideration to appeal general users.
  • Improving the co-dining experience for general users to reach sustainable development.
  • Taking eating habits as features to find those share the same habits and recommend foods through collaborative filtering.


Daily Machine Learning 100
  • ML Theory
  • Data Engineering
  • ML Algorithms
  • Deep Learning
  • Github Link

Skills & Knowledge


  • Machine Learning 
    • Classification, Regression, Regularization, Feature Engineering, Transfer Learning, Multi-task Learning.
    • Explainable Artificial Intelligence
      • Visualization
      • Influence Methods
      • Example-based Explanation
      • Evaluating Explanation
      • Human-involved Evaluation
  • Deep Learning 
    • RNN, Seq2Seq, Attention, Transformer.
  • Natural Language Processing
    • Natural Language Understanding
      • Natural Language Inference
      • Question Answering.
      • Span Detection.
    • Sentence Pair Modeling. 
    • Text Classification / Regression. 
    • Distributed word representation (word2vec / GloVe / FastText). 
    • Deep contextual representation (ELMO / BERT / GPT / XLNet / Roberta / DistilBERT / T5 ). 
  • ML Related Framework Experience 
    • TensorFlow 2.0, PyTorch, pytorch-transformers.
    • numpy, pandas, sklearn
  • Data Engineering Related Experience
    • Data collection - Amazon Mechanical Turk
    • RDBMS
  • UX/UI 
    • Design Thinking 
    • Prototyping 
    • Information Architecture 
    • Quantitative Research 
    • Qualitative Research 
    • User Definition

Powered by CakeResumePowered by CakeResume