Chenghong Wang

chwang.jpg

Assistant Professor
Department of Computer Science
Indiana University Bloomington
3054 Luddy Hall, Bloomington, IN 47408
Contact: cw166 AT iu (dOt) edu

I am an assistant professor in the Department of Computer Science at Indiana University. I am also affiliated with Luddy’s Security & Privacy in Informatics, Computing, and Engineering (SPICE) Center. I received my Ph.D. in Computer Science from Duke University under the supervision of Ashwin Machanavajjhala and Kartik Nayak.

I am looking for Ph.D. students and Postdocs to join our team. If any of my research topics capture your interest, eel free to reach out via email and include your CV.

Research

My current research focuses on designing secure-by-design, data-centric AI infrastructures that offer provable security and privacy, enabling a trustworthy foundation for modern AI applications. Some current research projects are:

  • Data infrastructures for trustworthy AI
    • ParsecDB (On-going): A highly efficient, full-fledged secure outsourced database that enables critical domain organizations to offload private data to the public cloud for analytics and ML ETL pipelines.
    • BOLT (On-going): A high-performance, secure, and data-oblivious KVS accelerator (the foundation for secure retrieval-based AI).
    • Picachv (Security25): A novel security monitor that formally and automatically enforces data use policies within data analytics and ML data pipelines.
    • DPAR (SC25): Private collective-communication AllReduce primitives for HPC-scale ML.
    • SPECIAL (VLDB24): The first secure workload planner designed for complex analytics over private data federations.
  • Learning-augumented private algorithms
    • DPidx: The first private learned index for secure outsourced databases and private data federations.
    • Security23: Use PAC-based learning tools, such as noisy binary trees and Bayesian learners, to quantify privacy leakages in PoS blockchains.
  • Confidential ML accelerators
    • LinGCN (NeurIPS23):An accelerator architecture designed to reduce multiplication depth of homomorphic encryption based GCN inference.
    • AQ2PNN (MICRO23): An ultra-fast 2PC-DNN accelerator built on FPGAs.

My research has received generous support from the NSF (2419821), IU IAS, Intel, and AMD

Students

  • Duo Xu
  • Weihong Sheng
  • Weijie Huang (with Haixu and XiaoFeng)
  • Jianzhang Du
  • Yitong Guo
  • Haobin Chen (with Haixu and XiaoFeng)
  • Hongbo Chen (with Haixu and XiaoFeng)
  • Leyi Zhao (Advisor Xuhong Zhang)

Alumni