About Me

Here is Lilian (Liyan Wang).
I am a student at Tianjin University, actively engaged in research on deep learning, bioinformatics, and multi-omics analysis for cancer. Currently, I am pursing a master degree in bioinformatics, advised by Prof. Kai Song. Here is my CV.

Academic Background

[!!] I am looking for Ph.D. position to start in 2024 Fall/2025 Spring. Contact me if you have any leads!

Sep 2021 -Now: M.E. in Chemical Process and Machinery, Tianjin University (Research Direction: Bioinformatics)
Sep 2017 - Jun 2021: B.S. in Processing Equipment and Control Engineering, Tianjin University (GPA: 3.6/4.0)

Research Interests

Bioinformatics

Enhanced Cross-Modal Fusion of Molecular and Image
Deep Learning of Genome-wide Features
Integrating Multimodal Biological Data
Deep learning-guided Radiomics Analysis

Machine Learning in Industry

Graph Neural Networks for Chemical Molecular Analysis
Simulation of Material Properties Optimized by Neural Networks
Chemical Process Fault Diagnosis
Crack Identification, SEM Image Analysis

My current research focuses on practical problems that artificial intelligence faces in medical and industry. My interests are on the Machine Learning and its applications in Medical and Industry.The vigorous development of deep learning technology has spawned a variety of interdisciplinary cross-applications. I hope that through unremitting efforts, I can make some contributions to this meaningful cause!

Research Records

Aug 2023: Preparing for the IELTS exam involves first completing fundamental vocabulary memorization and simultaneously studying reading and writing techniques. The anticipated schedule is to take the assessment by the end of September.Fighting!
Jul 2023: Customization of the contrastive loss function has been accomplished, alongside the construction of feature fusion networks for four types of omics data. After a week of optimization, fusion features of dimension 3*100 with substantial informative content have been obtained for survival analysis. Now, the process of drafting the research paper has commenced!
Jun 2023: Cleaning and initial feature screening of three types of omics data (gene expression, methylation, miRNA expression) from TCGAThrough statistical analysis, it has been found that cancer transcriptomic data commonly suffer from sample imbalance. After consulting the literature, two approaches have been identified for addressing this issue: 1) utilizing focal loss, and 2) employing SMOTETomek in the preprocessing stage. Considering the nature of the data, SMOTETomek was selected to achieve sample balancing.