Experiences
- Title: Statistical Modeling and Inference Based on Neural Network Prediction of Gene-Expression. Advisor: Tianying Wang.
- We applied the Mixture Density Network to predicting the gene expression level, studied the model performance and robustness;
- We extended the Conformal Prediction framework so as to apply to conctruct a conformal band for the distribution function of gene expression level.
- Studied landscape modification in Simulated Annealing, especially on discrete Hamiltonian, to speed up
sampling and optimization process;
- Focused on applicability of spin glass model and replica symmetric theory in explanation of landscape
modification.
- Crawled and parsed case-report articles on PubMed to form the PubMed-Center-Patient large-scale dataset
of Electronic Medical Record;
- Used PMC-Patient as seed dataset to fine-tune language model for crawling the whole PubMed OA;
- Mapped the citation graph as patients link as database for retrieval system;
- Crawled and parsed medical entity relation pairs from public medical websites to form knowledge graph.
- Worked on track A, concerning the modelling problem of cyclist stamina;
- Based on the simulation model, somatic function of cyclist could be estimated and random optimization was used to determine the best strategy for specific trial. Algorithm for team trial was also developed;
- Awarded Honorable Mention in MCM 2022.
- Studied on heterogenous junction between metal electrode and low-dimensional semiconductor MoS2 to
explore the characteristic and production technique;
- Different processing methods and techonologies were experimented to develop a better way, obtaining heterogenous junction with more stable and ideal performance.
- Further explored the usage of low-dimensional materials in ionic micro-device.