Murali Annavaram
Lloyd F. Hunt Chair of Electrical Power Engineering and Professor of Electrical and Computing Engineering and Computer Science
Builds systems for private and efficient machine learning, including secure and distributed training, along with computer architecture reliability.
Murali Annavaram's lab at USC works on efficient training & inference and accelerators, chips & architecture. 30 papers since 2023, 7 at top venues, mostly EMNLP and RecSys. 7 PhD students. 21 alumni, most often to NVIDIA and Google. Applications close .
At a glance
- papers, 2023–2026
- 8/ year
- at T1 venues
- 23%
- PhD students
- 7
- MS students
- 0
- alumni listed
- 21
Papers by yearfrom the lab homepage, 2023–2026
- 1 papers2026 · T1 1 · collecting
- 16 papers2025 · T1 3
- 9 papers2024 · T1 1
- 4 papers2023 · T1 2
T1 venuesOtherVenues: EMNLP, RecSys, ISCA, MICRO, WSDM
Research focuspaper titles that use each topic's words
- Efficient training & inference8 papers
- Accelerators, chips & architecture0 papers
- distributed training0 papers
- reliability0 papers
- efficient machine0 papers
Paper list
From the lab homepage.
20261 paper
- WSDM (T1)SAGERec: Sampling and Gating for Enhanced Long-Tail Item Recommendations
202516 papers
- Annual Conference of the Association for Computational Linguistics[ACL] C. Jiang*, L. Gao*, H. Zarch*, and M. Annavaram. KVPR: Efficient LLM Inference with I/O-Aware KV Cache Partial Recomputation. In 2025 Annual Conference of the Association for Computational Linguistics. July 2025
- Annual Conference of the Association for Computational Linguistics[ACL] J. Flemings*, B. Jiang, W. Zhang, Z. Takhirov, and M. Annavaram. Estimating Privacy Leakage of Augmented Contextual Knowledge in Language Models. In 2025 Annual Conference of the Association for Computational Linguistics. July 2025
- HyCache: Hybrid Caching for Accelerating Input Preprocessing Pipelines in DNN training, In USENIX ATC, June[ATC] K. Jha, S. Pandey, M. Annavaram, and A. Basu. HyCache: Hybrid Caching for Accelerating Input Preprocessing Pipelines in DNN training, In 2025 USENIX ATC, June 2025
- European Conference on Superconducting[EUCAS] T. Renduchintala*, J. Lee*, H., Zha*, M. Qi,* and M. Annavaram SuperCPU: A Decoupled Compute Paradigm towards a General-purpose Superconducting CPU Design
- DistilLock: Safeguarding LLMs from Unauthorized Knowledge Distillation on the Edge, NeurIPS Lock-LLM Workshop, Dec[LockLLM] L. Gao*, A. Mohanty, G. Kang, M. Annavaram. DistilLock: Safeguarding LLMs from Unauthorized Knowledge Distillation on the Edge, NeurIPS Lock-LLM Workshop, Dec 2025
- Annual Conference of the North American Chapter of the Association for Computational Linguistics[NAACL] A. Alsabnah*, and M. Annavaram. On Using Arabic Language Dialects in Recommendation Systems. In 2025 Annual Conference of the North American Chapter of the Association for Computational Linguistics. June 2025
- SEAS Workshop @ AAAI , Feb[SEAS] C. Jiang*, L. Gao*, M. Annavaram. Efficient LLM Inference with I/O-Aware Partial KV Cache Recomputation. SEAS Workshop @ AAAI 2025, Feb 2025
- Annual Conference on Language ModelingDEL: Context-Aware Dynamic Exit Layer for Efficient Self-Speculative Decoding
- PETSHigh-Throughput Secure Multiparty Computation with an Honest Majority in Various Network Settings
- RecSys (T1)LEAF: Lightweight, Efficient, Adaptive and Flexible Embedding for Large-Scale Recommendation Models
- proceedings of the annual Privacy Enhancing Technologies Symposium (PETS) JulyMeta-Learn to Unlearn: Enhanced Exact Machine Unlearning in Recommendation Systems with Meta-Learning
- EMNLP (T1)Mind the Dialect: NLP Advancements Uncover Fairness Disparities for Arabic Users in Recommendation Systems
20249 papers
- Annual Conference of the Association for Computational Linguistics[ACL] J. Flemings*, M. Annavaram. Differentially private knowledge distillation via synthetic text generation. In 2024 Annual Conference of the Association for Computational Linguistics. Aug 2024
- ENLSP workshop @ NeurIPS , Dec[ENLSP] L. Gao*, A. Ziashahabi. Y. Nue, S. Avestimehr, M. Annavaram. Enabling Resource-Efficient On-Device Fine-Tuning of LLMs Using Only Inference Engines. ENLSP workshop @ NeurIPS 2024, Dec 2024
- Annual Conference of the North American Chapter of the Association for Computational Linguistics[NAACL] J. Flemings*, M. Razaviyayn, M. Annavaram. Differentially private next token generation. In 2024 Annual Conference of the North American Chapter of the Association for Computational Linguistics. June 2024
- Annual Conference of the North American Chapter of the Association for Computational Linguistics[NAACL] L. Gao*, Y. Nie, T. Tang*, S. Avestimehr, M. Annavaram. Ethos: Rectifying Language Models in Orthogonal Parameter Space. In 2024 Annual Conference of the North American Chapter of the Association for Computational Linguistics. June 2024
- Accepted for PPAI workshop@AAAI , Feb[PPAI] J. Flemings*, M. Razaviyayn, M. Annavaram. Differentially Private Prediction of Large Language Models. Accepted for PPAI workshop@AAAI 2024, Feb 2024
- ReML workshop@AAAI , Spotlight Presentation, Feb[ReML] L. Gao*, Y. Niu, T. Tang*, S. Avestimehr, M. Annavaram. Ethos: Rectifying Language Models in Orthogonal Parameter Space. ReML workshop@AAAI 2024, Spotlight Presentation, Feb 2024
- Characterizing Context Memorization and Hallucination of Language Models, SATA Workshop @ NeurIPS , Dec[SATA] J. Flemings*, B. Ziang, Z. Takhirov, M. Annavaram. Characterizing Context Memorization and Hallucination of Language Models, SATA Workshop @ NeurIPS 2024, Dec 2024
- RecSys (T1)Biased User History Synthesis for Personalized Long-Tail Item Recommendation
- proceedings of the annual Privacy Enhancing Technologies Symposium (PETS) JulyEdge Private Graph Neural Networks with Singular Value Perturbation
20234 papers
- 48, 4617-4620[Optics] A. Minoofar, N, Karapetyan, A, Almaiman, H. Zhou, H. Song, K. Zou, W. Ko, M. Ramakrishnan, M. Annavaram, J.L. Habif, M. Tur, and A.E. Willner, Demonstration of optically-assisted reconfigurable average of two 20-Gbaud 4-phase-encoded data channels using nonlinear wave mixing. Opt. Lett. 48, 4617-4620 (2023)
- ISCA (T1)LAORAM: Look Ahead ORAM for Recommendation Model Privacy
- IEEE Transactions on Applied SuperconductivityResults from the ColdFlux Superconductor Integrated Circuit Design Tool Project
- MICRO (T1)SuperBP: Design Space Exploration for SFQ Branch Predictors
Where alumni went21 listed, 60% to industry
Industry 9Academia 6
Most often to
- NVIDIA
- Apple
Apply to USCFall 2027
Computer Science PhD (Viterbi)
- Round
- Single round (Dec 15)
- Opens
- —
- GRE
- See program page
- English
- See program page
- Fee
- See program page
Confirmed on the program page.
Similar labsnearest by research summary
- DETUM
Markus Götz
- Distributed, cloud & edge systems
- CNZhejiang
Kun Yang
- Accelerators, chips & architecture
- USUSC
Seo Jin Park
- Efficient training & inference
- SGNUS
Yang You
- Efficient training & inference
- USUT Austin
Diana Marculescu
- Accelerators, chips & architecture
- USUCSD
Farinaz Koushanfar
- Efficient training & inference
Your notes
Save this lab to keep notes.
Data
Checked against the lab homepage.
- Collected
- Verified
- Papers
- The lab homepage
- Members
- Lab homepage
- Alumni
- Lab homepage
Tiers are venue classifications, not lab rankings.
USCcloses