Geng Li

Postgraduate of Computer Science

Harbin Institute of Technology

Biography

Geng Li is a postgraduate student at Harbin Institute of Technology(HIT), Massive Data Computing(MDC) Lab. His research interests include few-shot learning, meta-learning, AutoML. He has two finished works focusing on meta-learning and AutoML, one is accepted by WISE 2022, the other is under review of TKDE(after first rebuttal). Two works are in progress, one work about architecture-specific meta-learning is in schedule for CVPR 2023, the Other work about meta-learning for Database Knobs Auto-tune is in schedule for ICDE 2023.

Download my RESUME/简历.

Interests

• Artificial Intelligence
• Few-shot Learning
• Meta-learning

Education

• Msc in Computer Science

  Harbin Institute of Technology

• Bsc in Computer Science

  Harbin Institute of Technology

Experience

 

 

 

 

 

Postgraduate Student

Massive Data Center Lab, HIT

Aug 2021 – Jun 2023 Harbin, China

Under the guidance of Hongzhi Wang, I am studying in Massive Data Center Lab, HIT, mainly focusing on few-shot learning and meta-learning algorithm innovation and application with database problems (AI4DB).

 

 

 

 

 

Algorithm Engineer Intern

JD.COM

Jul 2022 – Sep 2022 Beijing, China

As an algorithm engineer intern of geographical map algorithm group, logistics department, JD.COM. I mainly focus on solving Delivery Block Division problem, one of combinatorial optimization assignment problems with geographical constraints. Get praised as the best intern by department leader & mentor (oral:). Related works are concluded in Slides.

 

 

 

 

 

Exchange Visiting Student

Johns Hopkins University

Aug 2019 – Dec 2019 Baltimore, US

Under the exchange program between Johns Hopkins University and Harbin Institute of Technology (Honor School). The program covers the 2019 fall semester.

 

 

 

 

 

Undergraduate Student

Harbin Institute of Technology

Aug 2017 – Aug 2021 Harbin, China

Harbin Institute of Technology(HIT) is a public research university and a member of China’s elite C9 League. HIT is a Chinese Ministry of Education Class A Double First Class University. I graduated from the Honors School & faculty of computing.

Accomplish­ments

School Excellent Student Cadre

Harbin Institute of Technology Jan 2022 – Jul 2022

Competition: 23/1731 in Alibaba Cloud International Competition of Infrastructure Supply Chain

China Simulation Federation Nov 2021 – Jan 2022

Outstanding Graduates of University

Harbin Institute of Technology Jul 2021 – Dec 2021

Top Ten Outstanding Graduates of Honors School

Honors School, Harbin Institute of Technology Jul 2021 – Dec 2021

Competition: The Second Prize in 2020 National Collegiate Curling Artificial Intelligence Challenge

China Simulation Federation Jul 2020 – Dec 2020

National Encouragement Scholarship

Harbin Institute of Technology Jan 2020

The First Prize of the people’s scholarship in China

Harbin Institute of Technology Jan 2020

School Outstanding Student/Dean’s List

Faculty of Computing, HIT Jul 2019 – Dec 2020

National Encouragement Scholarship

Harbin Institute of Technology Jan 2019

The First Prize of the people’s scholarship in China

Harbin Institute of Technology Jan 2019

National Encouragement Scholarship

Harbin Institute of Technology Jan 2018

The First Prize of the people’s scholarship in China

Harbin Institute of Technology Jan 2018

School Outstanding Student/Dean’s List

Faculty of Computing, HIT Jul 2017 – Dec 2018

Projects

*

All Deep Learning Database Other

AI4DB: Quickly responding mechanism of validating knobs efficacy for GaussDB.

Exploiting meta-learning mechanism to enhance model generalization with dynamically changing workload of database, with the aim of decreasing model retraining cost and improving overall knobs evaluation performance.

opengauss webpage

Kaggle: BirdCLEF 2022 Identify bird calls in soundscapes.

Designed two phases (detect & classify) to identify birds species from sound segment; Applied Audio Spectrogram Transformer (AST), Vision Transformer (ViT), AlexNet, CNN-RNN network as backbone to both detect whether bird sound exists in segment and which kinds they are.

competition website

TIANCHI: 2021 Alibaba Cloud Infrastructure Supply Chain Competition.

Propose a two components (prediction & decision) solution to help Alibaba Cloud (cloud host provider) maintain best-profits supply chain; Utilized LSTM, ARIMA, DeepAR time-sequential models to predict future demands; Designed dynamic safety stock policy to support decision.

competition website

Course Project: Machine Learning during Visiting-Exchange in Johns Hopkins University.

Reproduced classical machine learning algorithm: Logistic Regression, Decision Tree(ID3), KNN, GMM, K-means, MLP, MRF, SVM and validated efficacy on public benchmarks.

course website

Publications

Geng Li, Boyuan Ren, Hongzhi Wang

July, 2022 Accepted by WISE 2022

EEML: Ensemble Embedded Meta-learning

This work is mainly about few-shot training paradigm of integrating ensemble learning with gradient-based meta-learning to utilize advantages of divergence and cooperation between experts to attain efficacy improvement.

PDF Cite DOI

Chunnan Wang, Bozhou Chen, Geng Li, Hongzhi Wang

July, 2022 under review of TKDE(after first rebuttal)

FL-AGCNS: Federated Learning Framework for Automatic Graph Convolutional Network Search

Federated learning graph convolutional network architecture search(fl-gcnas).

PDF Cite DOI