Projects

• Nov. 2018 - Mar. 2019: Student researcher for learning robust features for domain adaptation
  • Proposed a novel method for unsupervised domain adaptation by injecting target categorical knowledge directly into the model in an adversarial fashion.
  • Provided theoretical insights about why the proposed model enables fine-grained domain invariance within each category based on the relations between decision boundaries of classifiers and objective functions.
  • Implemented all the experiments in the proposed method and obtained improved performance against state-of-the-art transfer learning methods.
• Feb. 2018 - Oct. 2018: Student researcher for exploiting mutual relation for visual relationship detection
  • Intern at Center for Future Media, under the instruction of Prof. Yang Yang.
  • Co-proposed an effective algorithm for visual relationship detection by exploiting mutual relation in a siamese network and incorporating semantic information in the model.
• Mar. 2018 - Sept. 2018: Team member of a 3-people team in Undergraduates Innovation and Entrepreneurship Competition
  • Made contributions to design an anti-lost tracker with wireless communication chips for tracking monitored items.
  • Won a prize of 1500 RMB (top 15%).
• Sept. 2017 - Jan. 2018: Multi-Clue based Representation Learning for Doctors Clustering
  • Proposed a method to mitigate the problem of data sparsity for doctors clustering by utilizing various types of clue (e.g. rating number and comment text).
  • Crawled various types of data and built an auto-encoder for discriminative representation learning.
• Feb. 2018 - Jan. 2018:Assessment of Undergraduates’ Stress Level Based on Multi-Contributors Mining PDF
  • Evaluate the level of mental stress in undergraduates and analyse factors that contribute to it by principal component analysis and entropy weight method.
• July. 2017 - Sept. 2017 Intern at Center for Future Media, UESTC
  • Under the instruction of Prof. Yang Yang.
  • Finished Stanford Unsupervised Feature Learning and Deep Learning (Stanford-UFLDL) tutorial. Code
• Nov. 2017 - Feb. 2018 Mathematical Contest In Modeling/Interdisciplinary Contest In Modeling (MCM/ICM)
  Awarded “Honorable Mention”
  • Project 1: Evaluation on Climate-based Fragile State Index Project
  • Project 2: The Dynamic Path Planning of Drone Clusters Based on the Improved Artificial Potential Field Project
  • Project 3: Measuring the Evolution and Influence in Society’s Information Networks Project
  • Project 4: An Evaluation System for Smart Growth of a City Project
• Apr. 2018 - Jun. 2018 Fire Alarm Circuit Based on Temperature with Tunable Sensitivity (Chinese) PDF
  • Designed a circuit that keeps silent and provide no visual or acoustic signal in normal temperature. Once the temperature reachs an abnormal value, the circuit rings the buzzer and light the LED.
  • The sensitivity or the temperature threshold can be easily tuned by adjusting the variable resistor.
• Some Curriculum Designs
  • Course Algorithm Desgin and Analysis and ACM-ICPC contests holden by UESTC. Code