Elamur Research⌕
☰
home / authors / Han Cai

Han Cai

Han Cai has 4 papers in our library, in 2017–2020, most of them on GPU & parallel computing; often with Song Han, Chuang Gan, Ligeng Zhu.

4 papers in our library · 2017–2020

OpenAlex · ORCID · h-index 27 · 114 works · 4,801 citations OpenAlex, 2026-10-14

📄 4 works in our library · 🔗 Cited 674 times in our library · 7 more in our library without a page yet · ⭐ Paper Stars ranking: #1870

The profile is the library's match of the name; two people with one name are not always told apart.

Elsewhere: Google Scholar · Semantic Scholar · dblp · ACL Anthology

 Affiliations

🏢NvidiaSanta Clara · United States2024–2026 · 11 paperscurrent🏢NvidiaReading · United Kingdom2024–2026 · 8 paperscurrent🎓Maryland Quantum Materials CenterCollege Park · United Statescurrent🎓University of Maryland, College ParkCollege Park · United StatesEmploymentcurrent🎓Massachusetts Institute of TechnologyCambridge · United States2019–2024 · 20 papers🎓Shanghai Jiao Tong UniversityShanghai · China2016–2019 · 10 papers

From the bylines of their papers (OpenAlex, 2026-10-14), ORCID and Wikidata; institutions on papers are matched by machine. 2 more institutions on single bylines are not listed.

 Fields, topics and concepts

Of its works in our library, with the number of works on each.

📚 Shelves 5The field of science this work belongs to (OpenAlex fields).Computer Science · 4 worksArtificial Intelligence · 4 worksComputer Vision and Pattern Recognition · 4 worksArts and Humanities · 1 workLanguage and Linguistics · 1 work

🧭 Topics 1Research subjects and languages the work is about.GPU & parallel computing · 2 works

 Works by year

2020s

20201 paper

2010s

20191 paper20181 paper20171 paper

 Papers

The most cited first (OpenAlex and our library).

  1. Efficient Architecture Search by Network Transformation (2017 · 629 citations)
    Automated neural architecture search with reinforcement learning or evolution has matched human-designed models.
  2. ProxylessNAS: Direct Neural Architecture Search on Target Task and Hardware (2018 · 479 citations)
    Neural architecture search typically costs thousands of GPU hours. Differentiable variants such as DARTS and One-Shot are cheaper, but they keep the activations of every candidate operation in memory, so memory grows with the number of choices. As a result, earlier methods search on cheaper stand-ins, such as smaller…
  3. HAT: Hardware-Aware Transformers for Efficient Natural Language Processing (2020 · 294 citations)
    Transformer models for machine translation are too slow on constrained devices. For example, Transformer-Big needs about 13G FLOPs and roughly 20 seconds on a Raspberry Pi to translate a 30-word sentence. The authors make two observations: FLOPs poorly predict real latency, and the architecture that runs fastest…
  4. Once-for-All: Train One Network and Specialize it for Efficient Deployment (2019 · 252 citations)
    Deploying neural networks across many devices (phones, CPUs, GPUs, FPGAs, microcontrollers) and latency budgets usually means designing or searching a new architecture and training it from scratch for each target, so the cost grows with every new scenario.

All works

11 works of theirs in our library, the most cited first; 4 with a page here, the others link their original until we summarise them.

  1. Efficient Architecture Search by Network TransformationProceedings of the AAAI Conference on Artificial IntelligenceHan Cai, Tianyao Chen, Weinan Zhang, et al.2017 · 629 citations
  2. ProxylessNAS: Direct Neural Architecture Search on Target Task and HardwarearXiv:1812.00332Han Cai, Ligeng Zhu, Song Han2018 · 479 citations
  3. EfficientViT: Lightweight Multi-Scale Attention for High-Resolution Dense Prediction ↗IEEE/CVF International Conference on Computer Vision (ICCV) · no summary yetHan Cai et al.2023 · 363 citations
  4. HAT: Hardware-Aware Transformers for Efficient Natural Language ProcessingAnnual Meeting of the Association for Computational Linguistics (ACL)Hanrui Wang, Zhanghao Wu, Zhijian Liu, et al.2020 · 294 citations
  5. Once-for-All: Train One Network and Specialize it for Efficient DeploymentarXiv:1908.09791Han Cai, Chuang Gan, Tianzhe Wang, et al.2019 · 252 citations
  6. MAgent: A Many-Agent Reinforcement Learning Platform for Artificial Collective Intelligence ↗arXiv:1712.00600 · no summary yetLianmin Zheng et al.2017 · 158 citations
  7. Sparse VideoGen: Accelerating Video Diffusion Transformers with Spatial-Temporal Sparsity ↗arXiv:2502.01776 · no summary yetHaocheng Xi et al.2025 · 56 citations
  8. Sparse VideoGen2: Accelerate Video Generation with Sparse Attention via Semantic-Aware Permutation ↗arXiv:2505.18875 · no summary yetShuo Yang et al.2025 · 44 citations
  9. Cosmos 3: Omnimodal World Models for Physical AI ↗arXiv (Cornell University) · no summary yetNVIDIA et al.2026 · 19 citations
  10. Lightning OPD: Efficient Post-Training for Large Reasoning Models with Offline On-Policy Distillation ↗arXiv:2604.13010 · no summary yetYecheng Wu, Song Han, Han Cai2026 · 10 citations
  11. Scaling Vision Pre-Training to 4K Resolution ↗IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · no summary yetBaifeng Shi et al.2025 · 7 citations

 Co-authors here

10 co-authors with a page, the most papers together first.

Venues

Where their works here were published, the most works first, with the editions of those that have a page.

 Cited by

Papers of our library cite their works 674 times (of the 149,439 papers whose bibliographies we read); 18 citing papers with a page here →.

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/authors/han-cai/