About me

I am a Postdoc at the Department of Automation, Tsinghua University, advised by Prof. Yunhao Liu.

My research interest lies in Analytical Machine Learning, AI4Science and Embodied AI, Multimodal Large Models for AGI.

I love the philosophy of the world and the time, love the same feeling as Heraclitus:

δὶς ἐς τὸν αὐτὸν ποταμὸν οὐκ ἂν ἐμβαίης.

you cannot step into the same river twice

I am interested in machine intelligence and exploring the intelligence of human mind.

Research Persuit

I am devoting all myself to pursue the birth of Human intelligence. Only after this goal is achieved can I truly retire :).

Long way to go.

🚀 News

• Mar, 24: We are thrilled to announce attending the competition Tsinghua University Postdoctoral Innovation&Entrepreneurship Competition!

Experience

2024.03 - present, Postdoc at Department of Automation, Thinghua University, Beijing, China, mentored by Dr. Yunhao Liu.

• Developed algorithms for single cell sequencing analysis, such as ANN, GNN models
• UMAP (Python): For non-linear dimensionality reduction. t-SNE (Python/R): For visualizing high-dimensional data. Louvain/Leiden Algorithm: For community detection in large networks.

2023.07 - 2024.02, Postdoc, Michigan, US, mentored by Dr. Yuying Xie.

• Developed algorithms for single cell sequencing analysis, such as ANN, GNN models
• UMAP (Python): For non-linear dimensionality reduction. t-SNE (Python/R): For visualizing high-dimensional data. Louvain/Leiden Algorithm: For community detection in large networks.

2022.01 - 2023.02, Lucy Labs, New York, US, mentored by Jakub Rehor.

• Developed quantitative trading algorithms using various mathematical knowledge and programming skills.
• Utilized various traditional financial markets' quantitative strategies along with development of various trad- ing systems.
• Designed and implemented autonomous package for trading data fetching, data clean and data storage in Google Cloud BigQuery automatically.
• Implemented a Google Cloud BigQuery database system to store exchange trading data.
• Implemented workflow package for machine learning algorithms on AWS with Ray and Kubernetes.

2022.05 - 2022.11, Huawei, Santa Clara, US.

• Designed and implemented model structure. Scene graph generation module, vision transformer module for object bounding box prediction. Layout generation and image generation module. Image manipulation by scene graph.
• Data clean and preprocessing with Visual Genome dataset of 108,077 images, 3.8 million object instances, 2.3 million relationships and COCO dataset of 330,000 images, 1.5 million object instances.

First & Co-First Author Publications (Core Contribution)

* Equal Contribution

Xiaoyan Li*, Alyssa R Sanderson, Selett S Allen, Rebecca H Lahr. Tap water fingerprinting using a convolutional neural network built from images of the coffee-ring effect. Analyst 145 (4), 1511-1523. [code&data]

X Wang, W Wang, G Lowry, X Li, Y Guo, T Li. Preparation of palladized carbon nanotubes encapsulated iron composites: highly efficient dechlorination for trichloroethylene and low corrosion of nanoiron. Royal Society open science 5 (6), 172242

Kunli Liu, AKM Atique Ullah, Aniwat Juhong, Chia-Wei Yang, Cheng-You Yao, Xiaoyan Li, Harvey L Bumpers, Zhen Qiu, Xuefei Huang. Robust Synthesis of Targeting Glyco-Nanoparticles for Surface Enhanced Resonance Raman Based Image-Guided Tumor Surgery. Small Science 4 (5), 2300154

Preprint

Xiaoyan Li. Optimal environmental condition for contaminants separation by coffee-ring effect

Xiaoyan Li. CNN-Vision-transformer model for elements concentration estimation by coffee-ring effect residue patterns

Conferences

Xiaoyan Li. Attention as a Driving Force in Large Language Model Advancements. The 7th annual conference on Cognitive Computational Neuroscience

Awards

• A low cost and fast tap water quality test method: Coffee Ring Effect. Michigan State University, 2016/11, Second prize
• Single-cell analysis competition. Neural information processing systems (NeurIPS), 2021/11, Rank one in modality task
• Water quality monitoring and contaminants analysis with coffee-ring effect by machine learning. College of Engineering, Michigan State University, 2020/12, PhD dissertation completion fellowship
• A low cost and fast tap water quality test method: Coffee Ring Effect. Michigan State University, 2017/9, Lahr Award
  

  Teaching

  Teaching Assistant, Department of Automation, Tsinghua University, Beijing, China

  • 2024 Spring: Artificial Intelligence of Things(AIoT)
  Teaching Assistant, Lyman Briggs College, Michigan State University, East Lansing, Michigan, US
  • 2016 Fall: LB 171L: Introductory Chemistry Laboratory I
  • 2017 Spring: LB 172: Principles of Chemistry II
  • 2017 Fall: LB 171L: Introductory Chemistry Laboratory I
  • 2018 Spring: LB 172: Principles of Chemistry II
  • 2018 Fall: LB 171L: Introductory Chemistry Laboratory I
  • 2019 Spring: LB 172: Principles of Chemistry II
  • 2019 Fall: LB 171L: Introductory Chemistry Laboratory I
  • 2020 Spring: LB 172: Principles of Chemistry II
  • 2020 Fall: LB 171: Principles of Chemistry I
  • 2021 Spring: LB 172: Principles of Chemistry II
  • 2021 Fall: EGR 102 Introduction to Engineering Modeling
  • 2022 Spring: EGR 102 Introduction to Engineering Modeling

  Books

  From llm to AGI: where we are now and what we need to do next.

  Projects

  Tap Water Fingerprinting by CNN model with Coffee-Ring Effect

  Tech Stack: Deep learning, Pytorch, CNN, K-means clustering

  Built a CNN model to classify water samples nanochromatography pattern into 6 classes and achieved comparable accuracy as human clustering on water treatment method with 76.7 ± 3.0%accuracy.

  Comparison of Classical and CNN methods on Human Expression recognition

  Tech Stack: Pytorch, SVM, Gaussian Process Classifier, Logistic Regression

  Collaborated with teammates to build and tested Logistic Regression Classifier 60.1%, Multi-layer Perception Classifier 60.4%, SVM Classifier 60.2%, CNN Classifier 90.0% on the CK+ (1.7GB) and AffactNet (55GB) dataset with image gray-scale transformation and Gaussian Blur.

  Wireless Mesh Network Channel Assignment

  Tech Stack: Distributed system, Wireless mesh network

  Generated random graphs with 10 to 100 nodes and tested the FNI performance based on proposed algorithm and base algorithm. Helped to designed a scoring function to determine which channel to assign and the new model reduced the Fractional Net-work Interference (FNI) around 50\% than the base algorithm.

  Database Functions Implementation

  Tech Stack: SQLite, Wait-die/Wound-wait scheduler, loss-join, Conflict serializable

  Implemented a validation-locking schedule function to validate the schedule of the legality, two-phased locking, and consistency errors in the actions and the conflict-serializable function to verify the schedule is serializable. Implemented a transaction concurrency control scheduler based on wait die protocol.

  Academic Services

  Reviewer: CCN 2024, Tsinghua Science Technology, ACM Transactions on Sensor Networks (TOSN).

  Associate Editor: Tsinghua Science Technology

  Editor-in-Chief Assistant: Communications of the CCF

  Artificial Intelligence & Pattern Recognition: Specialist member

  Internet of Things: Specialist member

  Chinese association for Artificial Intelligence: Specialist member

  American Chemical Society: Specialist member

  Inspirational Persons

  Below are some persons from around the world whom I greatly admire:

  Andrej Karpathy, Mu Li: They have had a profound impact on me with their incredible open-source spirit.

  John Hopcroft, Andrew Yao,

  I had a personal conversation with Dr. John Hopcroft about our roles as research scientists and how to balance our research interests with quality pressures (2024 summer). I asked him if pursuing work that interests me, even if it ultimately fails, would be a bad choice. He told me that even if the research project doesn't succeed, I can still gain valuable and enjoyable experiences from the process, so there's nothing to lose.

  Edward Witten: He revolutionized my understanding of the physical world surrounding me.