About Me

I am a Computer Science Ph.D. candidate at Penn State University. My research interests lie in computer vision and MLLM. My recent work focuses on 3D vision and video generation.

Previously, I worked on semantic retrieval and image captioning.

News

  • [June 2026] Our paper “When 3D Gaussian Splatting Recovers Real Surfaces” was accepted to ECCV 2026.
  • [May 2024] Joined Intuit as a Research Scientist Engineer intern to train and deploy LLMs and VLMs.
  • [Mar. 2024] Our work on temporal-distributed backdoor attacks for video action recognition was published at AAAI 2024.
  • [Dec. 2023] Our work on backdoor threats from compromised foundation models appeared at FL@FM, NeurIPS 2023.
  • [Aug. 2021] Started my Ph.D. in Computer Science at Penn State University.

Publications

Selected publications. My name is shown in bold. * Equal contribution.

Figure from When 3D Gaussian Splatting Recovers Real SurfacesECCV

When 3D Gaussian Splatting Recovers Real Surfaces

Songhe Wang, David Jonathan Miller

ECCV, 2026.

Details available in CV

Figure from Temporal-Distributed Backdoor Attack Against Video Based Action RecognitionAAAI

Temporal-Distributed Backdoor Attack Against Video Based Action Recognition

Songhe Wang*, Xi Li*, Ruiquan Huang, Mahanth Gowda, George Kesidis

AAAI, 2024.

PaperarXiv
Figure from Backdoor Threats from Compromised Foundation Models to Federated LearningNeurIPSW

Backdoor Threats from Compromised Foundation Models to Federated Learning

Xi Li, Songhe Wang, Chen Wu, Hao Zhou, Jiaqi Wang

NeurIPSW, 2023.

PaperWorkshopCode
Figure from Armor: A Benchmark for Meta-evaluation of Artificial MusicACM MM

Armor: A Benchmark for Meta-evaluation of Artificial Music

Songhe Wang, Zheng Bao, Jingtong E

ACM MM, 2021.

PaperACM
Figure from A Spatial-Temporal Analysis of COVID-19's Impact on Human Mobility: The Case of the United StatesCICTP

A Spatial-Temporal Analysis of COVID-19's Impact on Human Mobility: The Case of the United States

Songhe Wang, Kangda Wei, Lei Lin, Weizi Li

CICTP, 2021.

PaperProceedings
Figure from Revealing the Importance of Semantic Retrieval for Machine Reading at ScaleEMNLP

Revealing the Importance of Semantic Retrieval for Machine Reading at Scale

Yixin Nie, Songhe Wang, Mohit Bansal

EMNLP, 2019.

PaperCode

Research Experience

Research Scientist, BoostDraft

LLM-powered multi-agent orchestration, legal document alignment, clustering, and term retrieval.

Research Scientist Engineer, Intuit

Training and deployment of LLMs and vision-language models for structured financial documents.

Research Assistant, Yale University

Cross-domain adaptation for text-to-SQL semantic parsers.