Lalitha Sankar
7 papers
· 2022–2025
· 4 conferences
· across top CS/AI conferences
Achievements
π
Interdisciplinary Bridge
π§
Keyword Pioneer
π
Conference Polyglot
(4)
π
Cross-Pollinator
(12)
πΊοΈ
Taxonomy Completionist
(10)
Conferences
ICML (4)
AISTATS (1)
ICLR (1)
NIPS (1)
Top co-authors
Research topics
Keywords
label noise
(2)
non-convex optimization
(1)
logistic regression
(1)
posterior distribution
(1)
boosting algorithm
(1)
loss function
(1)
proper loss
(1)
privacy accounting
(1)
margin-based loss
(1)
robust training
(1)
worst-group accuracy
(1)
subgroup fairness
(1)
saddle-point approximation
(1)
moments accountant
(1)
privacy loss distribution
(1)
last-layer retraining
(1)
alpha loss
(1)
saddle-point method
(1)
privacy curve
(1)
differential privacy
(1)
Papers
Smoothly Giving up: Robustness for Simple Models
AISTATS 2023
Being Properly Improper
ICML 2022