Publications
2024
- Enabling Visual Recognition at Radio FrequencyIn Proceedings of the 30th Annual International Conference on Mobile Computing and Networking , Washington D.C., DC, USA, 2024
This paper introduces PanoRadar, a novel RF imaging system that brings RF resolution close to that of LiDAR, while providing resilience against conditions challenging for optical signals. Our LiDAR-comparable 3D imaging results enable, for the first time, a variety of visual recognition tasks at radio frequency, including surface normal estimation, semantic segmentation, and object detection. PanoRadar utilizes a rotating single-chip mmWave radar, along with a combination of novel signal processing and machine learning algorithms, to create high-resolution 3D images of the surroundings. Our system accurately estimates robot motion, allowing for coherent imaging through a dense grid of synthetic antennas. It also exploits the high azimuth resolution to enhance elevation resolution using learning-based methods. Furthermore, PanoRadar tackles 3D learning via 2D convolutions and addresses challenges due to the unique characteristics of RF signals. Our results demonstrate PanoRadar’s robust performance across 12 buildings. Code, datasets, and demo videos are available on our website.
@inproceedings{10.1145/3636534.3649369, author = {Lai, Haowen and Luo, Gaoxiang and Liu, Yifei and Zhao, Mingmin}, title = {Enabling Visual Recognition at Radio Frequency}, year = {2024}, isbn = {9798400704895}, publisher = {Association for Computing Machinery}, address = {New York, NY, USA}, url = {https://doi.org/10.1145/3636534.3649369}, doi = {10.1145/3636534.3649369}, booktitle = {Proceedings of the 30th Annual International Conference on Mobile Computing and Networking}, pages = {388–403}, numpages = {16}, keywords = {RF sensing, mmWave radar, egomotion estimation, 3D imaging, robust perception, machine learning}, location = {Washington D.C., DC, USA}, series = {ACM MobiCom '24}, src="assets/img/publication_preview/acm_badge/artifacts_available_v1_1.png" alt="ACM Artifacts Available" style="width: 9em; height: auto;" > <img src="assets/img/publication_preview/acm_badge/artifacts_evaluated_functional_v1_1.png" alt="ACM Artifacts Available" style="width: 9em; height: auto;" > <img src="assets/img/publication_preview/acm_badge/artifacts_evaluated_reusable_v1_1.png" alt="ACM Artifacts Available" style="width: 9em; height: auto;" > <img src="assets/img/publication_preview/acm_badge/results_reproduced_v1_1.png" alt="ACM Artifacts Available" style="width: 9em; height: auto;" >}, }
- An in-depth evaluation of federated learning on biomedical natural language processing for information extractionnpj Digital Medicine, May 2024
Language models (LMs) such as BERT and GPT have revolutionized natural language processing (NLP). However, the medical field faces challenges in training LMs due to limited data access and privacy constraints imposed by regulations like the Health Insurance Portability and Accountability Act (HIPPA) and the General Data Protection Regulation (GDPR). Federated learning (FL) offers a decentralized solution that enables collaborative learning while ensuring data privacy. In this study, we evaluated FL on 2 biomedical NLP tasks encompassing 8 corpora using 6 LMs. Our results show that: (1) FL models consistently outperformed models trained on individual clients’ data and sometimes performed comparably with models trained with polled data; (2) with the fixed number of total data, FL models training with more clients produced inferior performance but pre-trained transformer-based models exhibited great resilience. (3) FL models significantly outperformed pre-trained LLMs with few-shot prompting.
@article{Peng2024, author = {Peng, Le and Luo, Gaoxiang and Zhou, Sicheng and Chen, Jiandong and Xu, Ziyue and Sun, Ju and Zhang, Rui}, title = {An in-depth evaluation of federated learning on biomedical natural language processing for information extraction}, journal = {npj Digital Medicine}, year = {2024}, month = may, day = {15}, volume = {7}, number = {1}, pages = {127}, issn = {2398-6352}, doi = {10.1038/s41746-024-01126-4}, url = {https://doi.org/10.1038/s41746-024-01126-4}, }
- SREE’24MetaMate: Large Language Model to the Rescue of Automated Data Extraction for Educational Systematic Reviews and Meta-analysesXue Wang, and Gaoxiang LuoMay 2024
Systematic reviews and meta-analyses are crucial for synthesizing evidence but are time-consuming and labor-intensive, especially during data extraction. To address this challenge, we developed MetaMate, an open-access web-based tool leveraging large language models (LLMs) for automated data extraction in educational systematic reviews and meta-analyses. MetaMate utilizes a hierarchical schema and divide-and-conquer approach in its extraction chain, and a from-global-to-local lens and example retriever in its verification chain. We evaluated MetaMate’s performance on 32 empirical studies, extracting 20 data elements related to participants and interventions. MetaMate achieved high precision, recall, and F1 scores, with performance comparable to human coders when benchmarked against an expert-defined gold standard. Notably, MetaMate demonstrated advanced mathematical reasoning and semantic comprehension, surpassing keyword-based approaches and avoiding common human errors. As the first LLM-powered data extraction tool designed specifically for educational research, MetaMate has the potential to significantly streamline the systematic review process and reduce time and effort for researchers. MetaMate is available at https://metamate.online.
@misc{wang_luo_2024, title = {MetaMate: Large Language Model to the Rescue of Automated Data Extraction for Educational Systematic Reviews and Meta-analyses}, url = {osf.io/preprints/edarxiv/wn3cd}, doi = {10.35542/osf.io/wn3cd}, publisher = {EdArXiv}, author = {Wang, Xue and Luo, Gaoxiang}, year = {2024}, month = may, }
2023
- Rethinking Transfer Learning for Medical Image ClassificationIn Proceedings of the 34th British Machine Vision Conference, , Aberdeen, UK, May 2023
Transfer learning (TL) from pretrained deep models is a standard practice in modern medical image classification (MIC). However, what levels of features to be reused are problem-dependent, and uniformly finetuning all layers of pretrained models may be suboptimal. This insight has partly motivated the recent differential TL strategies, such as TransFusion (TF) and layer-wise finetuning (LWFT), which treat the layers in the pretrained models differentially. In this paper, we add one more strategy into this family, called TruncatedTL, which reuses and finetunes appropriate bottom layers and directly discards the remaining layers. This yields not only superior MIC performance but also compact models for efficient inference, compared to other differential TL methods. Our code is available at: https://github.com/sun-umn/TTL
@inproceedings{peng2023rethinking, title = {Rethinking Transfer Learning for Medical Image Classification}, author = {Peng, Le and Liang, Hengyue and Luo, Gaoxiang and Li, Taihui and Sun, Ju}, year = {2023}, publisher = {British Machine Vision Association}, booktitle = {Proceedings of the 34th British Machine Vision Conference,}, series = {BMVC '23}, location = {Aberdeen, UK}, url = {https://papers.bmvc2023.org/0881.pdf}, }
- JMIAbility of artificial intelligence to identify self-reported race in chest x-ray using pixel intensity countsJohn Lee Burns, Zachary Zaiman, Jack Vanschaik, Gaoxiang Luo , and 9 more authorsJournal of Medical Imaging, May 2023
PurposePrior studies show convolutional neural networks predicting self-reported race using x-rays of chest, hand and spine, chest computed tomography, and mammogram. We seek an understanding of the mechanism that reveals race within x-ray images, investigating the possibility that race is not predicted using the physical structure in x-ray images but is embedded in the grayscale pixel intensities.ApproachRetrospective full year 2021, 298,827 AP/PA chest x-ray images from 3 academic health centers across the United States and MIMIC-CXR, labeled by self-reported race, were used in this study. The image structure is removed by summing the number of each grayscale value and scaling to percent per image (PPI). The resulting data are tested using multivariate analysis of variance (MANOVA) with Bonferroni multiple-comparison adjustment and class-balanced MANOVA. Machine learning (ML) feed-forward networks (FFN) and decision trees were built to predict race (binary Black or White and binary Black or other) using only grayscale value counts. Stratified analysis by body mass index, age, sex, gender, patient type, make/model of scanner, exposure, and kilovoltage peak setting was run to study the impact of these factors on race prediction following the same methodology.ResultsMANOVA rejects the null hypothesis that classes are the same with 95% confidence (F 7.38, P < 0.0001) and balanced MANOVA (F 2.02, P < 0.0001). The best FFN performance is limited [area under the receiver operating characteristic (AUROC) of 69.18%]. Gradient boosted trees predict self-reported race using grayscale PPI (AUROC 77.24%).ConclusionsWithin chest x-rays, pixel intensity value counts alone are statistically significant indicators and enough for ML classification tasks of patient self-reported race.
@article{10.1117/1.JMI.10.6.061106, author = {Burns, John Lee and Zaiman, Zachary and Vanschaik, Jack and Luo, Gaoxiang and Peng, Le and Price, Brandon and Mathias, Garric and Mittal, Vijay and Sagane, Akshay and Tignanelli, Christopher and Chakraborty, Sunandan and Gichoya, Judy W. and Purkayastha, Saptarshi}, title = {{Ability of artificial intelligence to identify self-reported race in chest x-ray using pixel intensity counts}}, volume = {10}, journal = {Journal of Medical Imaging}, number = {6}, publisher = {SPIE}, pages = {061106}, keywords = {machine learning, bias, population imaging, x-ray, Artificial intelligence, Data modeling, Chest imaging, Binary data, Brain-machine interfaces, X-rays, Machine learning, Decision trees, Education and training, Performance modeling}, year = {2023}, doi = {10.1117/1.JMI.10.6.061106}, url = {https://doi.org/10.1117/1.JMI.10.6.061106}, }
2022
- Evaluation of federated learning variations for COVID-19 diagnosis using chest radiographs from 42 US and European hospitalsJournal of the American Medical Informatics Association, Oct 2022
Federated learning (FL) allows multiple distributed data holders to collaboratively learn a shared model without data sharing. However, individual health system data are heterogeneous. “Personalized” FL variations have been developed to counter data heterogeneity, but few have been evaluated using real-world healthcare data. The purpose of this study is to investigate the performance of a single-site versus a 3-client federated model using a previously described Coronavirus Disease 19 (COVID-19) diagnostic model. Additionally, to investigate the effect of system heterogeneity, we evaluate the performance of 4 FL variations.We leverage a FL healthcare collaborative including data from 5 international healthcare systems (US and Europe) encompassing 42 hospitals. We implemented a COVID-19 computer vision diagnosis system using the Federated Averaging (FedAvg) algorithm implemented on Clara Train SDK 4.0. To study the effect of data heterogeneity, training data was pooled from 3 systems locally and federation was simulated. We compared a centralized/pooled model, versus FedAvg, and 3 personalized FL variations (FedProx, FedBN, and FedAMP).We observed comparable model performance with respect to internal validation (local model: AUROC 0.94 vs FedAvg: 0.95, P = .5) and improved model generalizability with the FedAvg model (P \< .05). When investigating the effects of model heterogeneity, we observed poor performance with FedAvg on internal validation as compared to personalized FL algorithms. FedAvg did have improved generalizability compared to personalized FL algorithms. On average, FedBN had the best rank performance on internal and external validation.FedAvg can significantly improve the generalization of the model compared to other personalization FL algorithms; however, at the cost of poor internal validity. Personalized FL may offer an opportunity to develop both internal and externally validated algorithms.
@article{10.1093/jamia/ocac188, author = {Peng, Le and Luo, Gaoxiang and Walker, Andrew and Zaiman, Zachary and Jones, Emma K and Gupta, Hemant and Kersten, Kristopher and Burns, John L and Harle, Christopher A and Magoc, Tanja and Shickel, Benjamin and Steenburg, Scott D and Loftus, Tyler and Melton, Genevieve B and Gichoya, Judy Wawira and Sun, Ju and Tignanelli, Christopher J}, title = {{Evaluation of federated learning variations for COVID-19 diagnosis using chest radiographs from 42 US and European hospitals}}, journal = {Journal of the American Medical Informatics Association}, volume = {30}, number = {1}, pages = {54-63}, year = {2022}, month = oct, issn = {1527-974X}, doi = {10.1093/jamia/ocac188}, url = {https://doi.org/10.1093/jamia/ocac188}, eprint = {https://academic.oup.com/jamia/article-pdf/30/1/54/47829262/ocac188.pdf}, }
- SAMCNet: Towards a Spatially Explainable AI Approach for Classifying MxIF Oncology DataIn Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining , Washington DC, USA, Oct 2022
The goal of spatially explainable artificial intelligence (AI) classification approach is to build a classifier to distinguish two classes (e.g., responder, non-responder) based on the their spatial arrangements (e.g., spatial interactions between different point categories) given multi-category point data from two classes. This problem is important for generating hypotheses towards discovering new immunotherapies for cancer treatment as well as for other applications in biomedical research and microbial ecology. This problem is challenging due to an exponential number of category subsets which may vary in the strength of their spatial interactions. Most prior efforts on using human selected spatial association measures may not be sufficient for capturing the relevant spatial interactions (e.g., surrounded by) which may be of biological significance. In addition, the related deep neural networks are limited to category pairs and do not explore larger subsets of point categories. To overcome these limitations, we propose a Spatial-interaction Aware Multi-Category deep neural Network (SAMCNet) architecture and contribute novel local reference frame characterization and point pair prioritization layers for spatially explainable classification. Experimental results on multiple cancer datasets (e.g., MxIF) show that the proposed architecture provides higher prediction accuracy over baseline methods. A real-world case study demonstrates that the proposed work discovers patterns that are missed by the existing methods and has the potential to inspire new scientific discoveries.
@inproceedings{10.1145/3534678.3539168, author = {Farhadloo, Majid and Molnar, Carl and Luo, Gaoxiang and Li, Yan and Shekhar, Shashi and Maus, Rachel L. and Markovic, Svetomir and Leontovich, Alexey and Moore, Raymond}, title = {SAMCNet: Towards a Spatially Explainable AI Approach for Classifying MxIF Oncology Data}, year = {2022}, isbn = {9781450393850}, publisher = {Association for Computing Machinery}, address = {New York, NY, USA}, url = {https://doi.org/10.1145/3534678.3539168}, doi = {10.1145/3534678.3539168}, booktitle = {Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining}, pages = {2860–2870}, numpages = {11}, keywords = {mxif, oncology, spatial interactions, spatially explainable classifier}, location = {Washington DC, USA}, series = {KDD '22}, }