Arman Akbari

PhD Candidate at Northeastern University · akbari.ar@northeastern.edu

prof_pic.png

Northeastern University, Boston, MA

I am a PhD candidate in Computer Engineering at Northeastern University, co-advised by Prof. Yanzhi Wang and Prof. Xuan Zhang. I earned my M.S. in Computer Engineering from Northeastern University in 2026 and my B.Sc. in Computer Science from the University of Tehran in 2024.

My research spans two directions:

  • Efficient Physical AI — world action models and vision-language-action (VLA) models, with a focus on making them efficient enough for real-time, real-world deployment. I have hands-on experience across different platforms, including humanoids (Unitree G1, AgiBot) and robotic arms (UR3).
  • Privacy-Preserving Machine Learning — privacy and robustness of multimodal models and LLMs, approached through the lens of efficiency.

Feel free to reach out via email if you’d like to discuss research or potential collaborations.

News

Sep 24, 2026 ActQuant and Palette accepted in NeurIPS 2026!
Aug 31, 2026 Joined RoboForce logo RoboForce as an AI Resident.
May 21, 2026 Our paper Cross-Platform Scaling of Vision-Language-Action Models from Edge to Cloud GPUs was accepted at GLSVLSI 2026.
Apr 01, 2026 I received my M.S. in Computer Engineering from Northeastern University, with a concentration in Computer Vision, Machine Learning, and Algorithms (CVLA), along the way to my Ph.D.
Mar 23, 2026 Two papers were accepted at DAC 2026!
Jan 26, 2026 CircuitSense (full version) got accepted in ICLR 2026!
Oct 17, 2025 CircuitSense got accepted in NeurIPS 2025 MATH-AI, The 5th Workshop on Mathematical Reasoning and AI and will be presented in San Diego Convention Center on December 6th.
Sep 30, 2025 We release our paper Beyond Overall Accuracy: Pose- and Occlusion-driven Fairness Analysis in Pedestrian Detection for Autonomous Driving on ArXiv.
Sep 29, 2025 We arxived CircuitSense and the hierarchical synthetic generation code.
Jul 13, 2025 Our paper VOTE: Vision-Language-Action Optimization with Trajectory Ensemble Voting was accepted at ICANN 2026.
Sep 13, 2024 I am starting my PhD at Northeastern University.

Selected Publications * Equal Contribution

  1. Preprint
    Flash-WAM: Modality-Aware Distillation for World Action Models
    Arman Akbari, Ci Zhang, Arash Akbari, Lin Zhao, Yixiao Chen, Weiwei Chen, Xuan Zhang, Geng Yuan, and Yanzhi Wang
    arXiv preprint arXiv:2606.05254, 2026
    PDF Website
  2. Preprint
    Q-WAM: 4-Bit Quantization of World Action Models with Action-Subspace Protection
    Arash Akbari*, Arman Akbari*, Jingwu Luo, Yuhao Lei, Yi Gao, Weiwei Chen, Xuan Zhang, Zhenman Fang, Geng Yuan, and Yanzhi Wang
    arXiv preprint arXiv:2609.33269, 2026
    PDF
  3. NeurIPS
    ActQuant: Sub-4-bit Action-Guided Quantization for Vision-Language-Action Models
    Arash Akbari, Arman Akbari, Masih Eskandar, Qitao Tan, Yixiao Chen, Jingwu Luo, Bertha Pangaribuan, Liyun Zhang, Jennifer Dy, Geng Yuan, Xue Lin, Gaowen Liu, Stratis Ioannidis, and Yanzhi Wang
    In Advances in Neural Information Processing Systems (NeurIPS), 2026
    PDF Website
  4. NeurIPS
    Palette: A Modular, Controllable, and Efficient Framework for On-demand Authorized Safety Alignment Relaxation in LLMs
    Qitao Tan, Xiaoying Song, Arman Akbari, Arash Akbari, Yanzhi Wang, Xiaoming Zhai, Lingzi Hong, Zhen Xiang, Jin Lu, and Geng Yuan
    In Advances in Neural Information Processing Systems (NeurIPS), 2026
    PDF Code
  5. ICLR
    CircuitSense: A Hierarchical Circuit System Benchmark Bridging Visual Comprehension and Symbolic Reasoning in Engineering Design Process
    Arman Akbari, Jian Gao, Yifei Zou, Mei Yang, Jinru Duan, Dmitrii Torbunov, Yanzhi Wang, Yihui Ren, and Xuan Zhang
    In International Conference on Learning Representations (ICLR), 2026
    PDF Code
  6. DAC
    HoloCode: Hybrid Optical-Electronic Edge Encoding for Privacy-Preserving Cloud Training
    Ruofan Xing, Arman Akbari, Weikai Lin, Adith Boloor, Alexander McNeil, Michael Moebius, Yongmin Liu, Yuhao Zhu, and Xuan Zhang
    In ACM/IEEE Design Automation Conference (DAC), 2026
    HTML
  7. ICANN
    VOTE: Vision-Language-Action Optimization with Trajectory Ensemble Voting
    Juyi Lin, Amir Taherin, Arash Akbari, Arman Akbari, Lei Lu, Guangyu Chen, Taskin Padir, Xiaomeng Yang, Weiwei Chen, Yiqian Li, Xue Lin, David Kaeli, Pu Zhao, and Yanzhi Wang
    In International Conference on Artificial Neural Networks (ICANN), 2026
    PDF Code