Sanghamitra Dutta
Assistant Professor
Develops trustworthy machine learning using information theory, including fairness, explainability and counterfactual reasoning.
Sanghamitra Dutta's lab at UMD works on trustworthy & safe ML, learning theory & optimization and interpretability, evaluation & safety of LMs. 23 papers since 2023, 14 at top venues, mostly NeurIPS and TMLR. 7 PhD students and 4 master's students. 1 alumnus, most often to Capitalone AI Research.
At a glance
- papers, 2023–2026
- 6/ year
- at T1 venues
- 61%
- PhD students
- 7
- MS students
- 4
- alumni listed
- 1
Papers by yearfrom the lab homepage, 2023–2026
- 4 papers2026 · T1 2 · collecting
- 8 papers2025 · T1 6
- 5 papers2024 · T1 3
- 6 papers2023 · T1 3
T1 venuesOtherVenues: NeurIPS, TMLR, AAAI/ACM AIES, ICML, ISIT
Research focuspaper titles that use each topic's words
- Trustworthy & safe ML4 papers
- Learning theory & optimization2 papers
- explainability1 paper
- Interpretability, evaluation & safety of LMs0 papers
- trustworthy machine0 papers
Paper list
From the lab homepage.
20264 papers
- JSAITFoundations of Efficient Model Reconstruction Using Counterfactual Explanations
- NeurIPS (T1)Improving Consistency in Retrieval-Augmented Systems with Group Similarity Rewards
- NeurIPS (T1)LogT: Logically Think with Images for Visual Search
- IEEE Transactions on Information TheoryWhat If, But Privately: Private Counterfactual Retrieval
20258 papers
- AAMAS (T1)Counterfactual Explanations for Model Ensembles Using Entropic Risk Measures
- NeurIPS (T1)Few-Shot Knowledge Distillation of LLMs With Counterfactual Explanations
- ISITPrivate Counterfactual Retrieval With Immutable Features
- AISTATS (T1)Quantifying Knowledge Distillation using Partial Information Decomposition
- ICML (T1)Quantifying Prediction Consistency Under Fine-tuning Multiplicity in Tabular LLMs
- NeurIPS (T1)T-Shirt: Token-Selective Hierarchical Data Selection for Instruction Tuning
- TMLR (T1)Towards Formalizing Spuriousness of Biased Datasets Using Partial Information Decomposition
- AAAI/ACM AIESVISION: Robust and Interpretable Code Vulnerability Detection Leveraging Counterfactual Augmentation
20245 papers
- ISITA Unified View of Group Fairness Tradeoffs Using Partial Information Decomposition
- ICLR (T1)Demystifying Local & Global Fairness Trade-offs in Federated Learning Using Partial Information Decomposition
- TMLR (T1)Hyper-parameter Tuning for Fair Classification without Sensitive Attribute Access
- NeurIPS (T1)Model Reconstruction Using Counterfactual Explanations: A Perspective From Polytope Theory
- JSAITRobust Algorithmic Recourse Under Model Multiplicity with Probabilistic Guarantees
20236 papers
- EntropyA Review of Partial Information Decomposition in Algorithmic Fairness and Explainability
- FAccTCan Querying for Bias Leak Protected Attributes? Achieving Privacy With Smooth Sensitivity
- TMLR (T1)Fairness via In-Processing in the Over-parameterized Regime: A Cautionary Tale
- UAI (T1)In- or Out-of-Distribution Detection via Dual Divergence Estimation
- AAAI/ACM AIESREFRESH: Responsible and Efficient Feature Reselection guided by SHAP values
- ICML (T1)Robust Counterfactual Explanations for Neural Networks with Probabilistic Guarantees
Where alumni went1 listed, 100% to industry
Industry 1Academia 0
Most often to
- Capitalone AI Research
Apply to UMDFall 2027
Not announced yet
UMD Computer Science PhD
- Round
- Single round
- Opens
- —
- GRE
- See program page
- English
- See program page
- Fee
- See program page
Not yet confirmed for this intake.
Similar labsnearest by research summary
- USCMU
Nihar Shah
- Learning theory & optimization
- CAToronto
Konstantinos N. Plataniotis
- Signal processing & information theory
- CNTsinghua
Yujiu Yang
- Learning theory & optimization
- GBImperial
Lucia Specia
- Multilingual, translation & linguistics
- CNSJTU
Yibo Yang
- Trustworthy & safe ML
- USUSC
Sai Praneeth Karimireddy
- Learning theory & optimization
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.