*

Lianfa Li, PhD

**** *. **** ******

Los Angeles, California 90032

Phone: 650-***-**** Email: adzeuq@r.postjobfree.com

SUMMARY OF QUALIFICATIONS

• Over 8 years developing AI solutions for geospatial applications using satellite, aerial, and geospatial data sets.

• 20+ peer-reviewed publications on DL/ML algorithms, intelligent information extraction by remotely sensed data, spatial statistical, data analysis and applications

• Expertise in developing novel deep learning approaches like graph neural network, visual transformers and reinforcement learning for remote sensing challenges

• Led projects applying computer vision to satellite imagery for applications in land cove assessment and aerosol parameter inversion

• Skilled in rapid prototyping of novel algorithms, training optimized models and transitional research code into production environments

• Proficient in PyTorch, TensorFlow, Scikit-Learn, Pandas, R, GDAL,Bash, Git, Docker/PodMan, AWS/GCP. SELECTED RECENT PROJECTS

• Semantic Segmentation for Land Cover Mapping - Developed the novel techniques to optimize training samples and fine-tune large pre-trained models, significantly improving mean IoU by about 5-10% on land cover segmentation from satellite imagery. Enabled more accurate land use monitoring.

• Air Quality Assessment with Sparse Data - Created knowledge-informed deep (graph) learning models leveraging domain knowledge to enhance few-shot and transfer learning. Reduced bias by 11-42% compared to baseline models, allowing air quality prediction from limited sensors or sparse measurement data.

• Aerosol Data Imputation - Developed a residual deep learning approach to accurately impute missing values in atmospheric aerosol satellite data. Improved MAIAC AOD data completeness by over 40% for enhanced usability. PROFESSIONAL EXPERIENCE

Lead Data Scientist - University of Southern California (USC) Aug 2014 – Present

• Lead R&D applying AI and data science to extract insights from satellite, spatial, and air quality data sets

• Develop and deploy DL pipelines for land cover mapping, climate modeling, air quality monitoring

• Publish 15+ papers on novel techniques for RS analysis using computer vision and geospatial data science

• Guide team of domain experts and analysts to apply state-of-the-art ML to address critical challenges ML/DL Researcher, Remote Sensing - Spatial Data Intelligence Lab (part-time, remote) Aug 2020 – Present

• Develop novel DL approaches for land cover segmentation and object detection in satellite, aerial and UAV data

• Fine-tune transformers and diffusion models on remote sensing data, improving mean IoU by 5-10%

• Provide consulting on computer vision and geospatial data science for satellite, hyperspectral and UAV data EDUCATION

• PhD, Geographical Information Science – IGSNRR, Chinese Academy of Sciences, China 2008

• BSc, Resources, Planning and Management - Nanjing University, China 2001 SELECTED PEER-REVIEWED PUBLICATIONS

h-index: 31 (Google scholar)

1. Li, L., Wang, J., Franklin, M., et al. 2023, Improving air quality assessment using physics-inspired deep graph learning, npj Climate and Atmospheric Science (in press, IF: 9.4). 2

2. Li, L., Zhu, Z. & Wang, C. 2023, Multiscale entropy-based surface complexity analysis for land cover image semantic segmentation, Remote Sensing (IF: 5.00), 15, 2192. 3. Li, L., 2021, Deep learning: principles and remote sensing geoscience analysis (in Chinese), Science Press, https://github.com/lspatial/deeplearning_geoscience. 4. Li, L.; Fang, Y.; Wu, J. et al. 2021, Encoder–decoder full residual deep networks for robust regression and spatiotemporal estimation, IEEE Transactions on Neural Networks and Learning Systems (IF: 14.25), 2021, 32(9): 4217-423 5. Li, L.; Wu, J.2021, Spatiotemporal estimation of satellite-borne and ground-level NO2 using full residual deep networks, Remote Sensing of Environment (IF: 13.71), 254: 0-112257

6. Li, L., Franklin, M., Girguis, M. et al. 2020. Spatiotemporal imputation of MAIAC AOD using deep learning with downscaling. Remote Sensing of Environment (IF: 10.84), 237, p.111584. 7. Li, L., Girguis, M., Lurmann, F. et al., 2020. Ensemble-based deep learning for estimating PM2. 5 over California with multisource big data including wildfire smoke. Environment International (IF: 10.00), 145, p.106143. 8. Li, L., 2020, Deep residual autoencoder with multiscaling for semantic segmentation of land-use images, Remote Sensing (IF: 4.8), 11(18): 2142.

SELECTED SOFTWARE

1. Li, L. et al. PyTorch package for physics-aware deep graph learning for fine-scale air quality assessment, https://pypi.org/project/phygeograph/

• Implements methods of “Improving air quality assessment using physics-inspired deep graph learning”. 2. Li, L., Python library for bagging of deep residual neural networks: https://pypi.org/project/baggingrnet/

• Implements methods of Li, L., Girguis, M., Lurmann, F., et al. 2020, Environ. Int., 145, p. 106143. 3. Li, L., Python library for deep residual multiscale segmenter: https://pypi.org/project/resmcseg/

• Implements methods of Wang, C. & Li, L.,2021 Remote Sens., 12(18), p. 2932. and Li, L., 2019, Remote Sens., 11(18), p. 2142

4. Li, L., Python library of autoencoder-based residual deep network: https://pypi.org/project/resautonet/

• Implements methods of Li, L., Fang, Y., Wu, J., et al. 2021, IEEE Trans. Neural Netw. Learn. Syst, 32(9), pp. 4217-4230.

WORKSHOP AND PRESENTATION

1. Invited presentation: “GCN-assisted U-Net for segmentation of OCT images”, the Bay area data science workshop, Mar. 27, 2021.

2. Invited presentation: “Enhancing semantic segmentation with contextual information”, the Bay area data science workshop, Dec. 07, 2019.

3. Invited presentation: “Knowledge and big data modeling driven by stochastic algorithms” Guizhou Doctoral Forum, Apr. 21, 2018, Guiyang, Guizhou, China

4. Invited lecturer: “GWC artificial intelligence open course” Beijing, China, Apr. 28-29, 2017.

Contact this candidate