Farzan Farnia
Assistant Professor
Studies machine learning theory, deep generative models, optimization and information theory, including optimal transport.
Farzan Farnia's lab at CUHK works on learning theory & optimization, generative & foundation models and planning, control & multi-robot systems. 47 papers since 2023, 33 at top venues, mostly ICML and NeurIPS. Applications close .
At a glance
- papers, 2023–2026
- 12/ year
- at T1 venues
- 70%
- PhD students
- —
- MS students
- —
- alumni listed
- —
Papers by yearfrom DBLP, 2023–2026
- 7 papers2026 · T1 2 · collecting
- 22 papers2025 · T1 17
- 11 papers2024 · T1 8
- 7 papers2023 · T1 6
T1 venuesOtherVenues: ICML, NeurIPS, AISTATS, ICLR, UAI
Research focuspaper titles that use each topic's words
- Generative & foundation models14 papers
- Learning theory & optimization9 papers
- Planning, control & multi-robot systems2 papers
- optimal1 paper
- optimal transport0 papers
Paper list
From DBLP.
20267 papers
- AAAI (T1)Boosting Cross-problem Generalization in Diffusion-Based Neural Combinatorial Solver via Inference Time Adaptation (opens in a new tab)
- ICLR (T1)DAK-UCB: Diversity-Aware Prompt Routing for LLMs and Generative Models (opens in a new tab)
- DATEDiffResist: Physics-Constrained Diffusion for Photoresist Modeling (opens in a new tab)
- DATEFastRW: An Efficient Random Walk Method for Steady-State Thermal Analysis (opens in a new tab)
- Trans. Mach. Learn. Res.NeMoS: Nearest Neighbors Bandit meets Active Learning for Online Model Selection (opens in a new tab)
- ISITOn the Fragility of AI-Based Channel Decoders under Small Channel Perturbations (opens in a new tab)
- ISITThe Maximum von Neumann Entropy Principle: Theory and Applications in Machine Learning (opens in a new tab)
202522 papers
- AISTATS (T1)A Multi-Armed Bandit Approach to Online Selection and Evaluation of Generative Models (opens in a new tab)
- ICCVWAn Information-Theoretic Approach to Diversity Evaluation of Prompt-Based Generative Models (opens in a new tab)
- NeurIPS (T1)APOLLO: Automated LLM and Lean Collaboration for Advanced Formal Reasoning (opens in a new tab)
- ICLR (T1)Be More Diverse than the Most Diverse: Optimal Mixtures of Generative Models via Mixture-UCB Bandit Algorithms (opens in a new tab)
- ICLR (T1)Boosting the visual interpretability of CLIP via adversarial fine-tuning (opens in a new tab)
- ICML (T1)Certifiably Robust Model Evaluation in Federated Learning under Meta-Distributional Shifts (opens in a new tab)
- IEEE Trans. Comput. Aided Des. Integr. Circuits Syst.DiffPattern-Flex: Efficient Layout Pattern Generation via Discrete Diffusion (opens in a new tab)
- UAI (T1)Do Vendi Scores Converge with Finite Samples? Truncated Vendi Score for Finite-Sample Convergence Guarantees (opens in a new tab)
- AISTATS (T1)Gaussian Smoothing in Saliency Maps: The Stability-Fidelity Trade-Off in Neural Network Interpretability (opens in a new tab)
- ICML (T1)Kernel-based Unsupervised Embedding Alignment for Enhanced Visual Representation in Vision-language Models (opens in a new tab)
- NeurIPS (T1)miniF2F-Lean Revisited: Reviewing Limitations and Charting a Path Forward (opens in a new tab)
- ICML (T1)Multilayer Matrix Factorization via Dimension-Reducing Diffusion Variational Inference (opens in a new tab)
202411 papers
- BMVCA Super-pixel-based Approach to the Stable Interpretation of Neural Networks (opens in a new tab)
- ICML (T1)An Information Theoretic Approach to Interaction-Grounded Learning (opens in a new tab)
- ICML (T1)An Interpretable Evaluation of Entropy-based Novelty of Generative Models (opens in a new tab)
- DAC (T1)ChatPattern: Layout Pattern Customization via Natural Language (opens in a new tab)
- AISTATS (T1)On Convergence in Wasserstein Distance and f-divergence Minimization Problems (opens in a new tab)
- UAI (T1)On the Inductive Biases of Demographic Parity-based Fair Learning Algorithms (opens in a new tab)
- ICLR (T1)Provably Efficient CVaR RL in Low-rank MDPs (opens in a new tab)
- ACCVSparse Domain Transfer via Elastic Net Regularization (opens in a new tab)
- Trans. Mach. Learn. Res.Stability and Generalization in Free Adversarial Training (opens in a new tab)
- CVPR (T1)Structured Gradient-Based Interpretations via Norm-Regularized Adversarial Training (opens in a new tab)
- NeurIPS (T1)Towards a Scalable Reference-Free Evaluation of Generative Models (opens in a new tab)
20237 papers
- NeurIPS (T1)An Information-Theoretic Evaluation of Generative Models in Learning Multi-modal Distributions (opens in a new tab)
- DAC (T1)DiffPattern: Layout Pattern Generation via Discrete Diffusion (opens in a new tab)
- SIAM J. Math. Data Sci.GAT-GMM: Generative Adversarial Training for Gaussian Mixture Models (opens in a new tab)
- ICASSP (T1)Interpretation of Neural Networks is Susceptible to Universal Adversarial Perturbations (opens in a new tab)
- AISTATS (T1)Mode-Seeking Divergences: Theory and Applications to GANs (opens in a new tab)
- ICCV (T1)MoreauGrad: Sparse and Robust Interpretation of Neural Networks via Moreau Envelope (opens in a new tab)
- UAI (T1)On the Role of Generalization in Transferability of Adversarial Examples (opens in a new tab)
Where alumni went
Alumni are not listed on the lab homepage.
Apply to CUHKFall 2027
Computer Science and Engineering MPhil/PhD
- Round
- HKPFS / main round
- Opens
- —
- GRE
- See program page
- English
- Interview in English
- Fee
- See program page
Not yet confirmed for this intake.
Similar labsnearest by research summary
- CNTsinghua
Xiu Li
- Reinforcement learning
- CNZhejiang
Zunlei Feng
- Medical image analysis
- USNYU
Yann LeCun
- Generative & foundation models
- USUC Berkeley
Laura Waller
- Computational imaging & photography
- USUW
Linda Shapiro
- Medical image analysis
- DETUM
Rüdiger Westermann
- Rendering & appearance
Your notes
Save this lab to keep notes.
Data
Some numbers could not be verified:
- The member list could not be found on the lab site.
- The alumni list could not be found on the lab site.
- Collected
- Verified
- Papers
- DBLP
- Members
- Not collected
- Alumni
- Not collected
Tiers are venue classifications, not lab rankings.
CUHKcloses