Scholar
Lun Wang
Google Scholar ID: _TkfqdgAAAAJ
Google Deepmind
Citations & Impact
Citations
1,763
H-index
16
i10-index
20
Publications
20
Co-authors
17
list available
Publications
Resume (English only)
Academic Achievements
- Attacker's Noise Can Manipulate Your Audio-based LLM in the Real World.
- Can DeepFake Speech be Reliably Detected?
- Differentially Private Parameter-Efficient Fine-tuning for Large ASR Models.
- Interspeech 2025: The 26th edition of the Interspeech Conference.
- Revisit Micro-batch Clipping: Adaptive Data Pruning via Gradient Manipulation.
- AudioMarkBench: Benchmarking Robustness of Audio Watermarking.
- Efficiently Train ASR Models that Memorize Less and Perform Better with Per-core Clipping.
- AITIA: Efficient Secure Computation of Bivariate Causal Discovery.
- Unintended Memorization in Large ASR Models, and How to Mitigate It.
- Why Is Public Pretraining Necessary for Private Model Training?
- Secure Federated Correlation Test and Entropy Estimation.
- Byzantine-Robust Federated Learning with Optimal Rates and Privacy Guarantee.
- Differentially Private Fractional Frequency Moments Estimation with Polylogarithmic Space.
- PRIVGUARD: Privacy Regulation Compliance Made Easier.
- BACKDOORL: Backdoor Attack against Competitive Reinforcement Learning.
- Towards practical differentially private causal graph discovery.
- Towards Inspecting and Eliminating Trojan Backdoors in Deep Neural Networks.
Research Experience
- Serves as a research scientist at Google Deepmind, focusing on the Gemini project.
Education
- Received a PhD in Computer Science from UC Berkeley in summer 2022, advised by Prof. Dawn Song; obtained a Bachelor's degree with honors in Computer Science from Peking University in fall 2018.
Background
- A staff research scientist at Google Deepmind, working on Gemini post-training (Memory, Tool Use, and Audio). Main research interests lie in computer science.
