activity
20102022
most citedLearning Topic Models - Going beyond SVD

60 citations · 204 across the 15 of their papers we have counts for

collaborators
Showing 2022Show all

6 papers · 1 filter

cs.DS2022

Minimax Rates for Robust Community Detection

Allen Liu, Ankur Moitra

In this work, we study the problem of community detection in the stochastic block model with adversarial node corruptions. Our main result is an efficient algorithm that can tolera…

cs.DS2022

From algorithms to connectivity and back: finding a giant component in random k-SAT

Zongchen Chen, Nitya Mani, Ankur Moitra

We take an algorithmic approach to studying the solution space geometry of relatively sparse random and bounded degree -CNFs for large . In the course of doing so, we establi…

cs.LG2022★ 3 cited

Learning in Observable POMDPs, without Computationally Intractable Oracles

Noah Golowich, Ankur Moitra, Dhruv Rohatgi

Much of reinforcement learning theory is built on top of oracles that are computationally hard to implement. Specifically for learning near-optimal policies in Partially Observable…

stat.ML2022

Provably Auditing Ordinary Least Squares in Low Dimensions

Ankur Moitra, Dhruv Rohatgi

Measuring the stability of conclusions derived from Ordinary Least Squares linear regression is critically important, but most metrics either only measure local stability (i.e. aga…

cs.LG2022★ 11 cited

Distilling Model Failures as Directions in Latent Space

Saachi Jain, Hannah Lawrence, Ankur Moitra +1

Existing methods for isolating hard subpopulations and spurious correlations in datasets often require human intervention. This can make these methods labor-intensive and dataset-s…

cs.LG2022★ 4 cited

Planning in Observable POMDPs in Quasipolynomial Time

Noah Golowich, Ankur Moitra, Dhruv Rohatgi

Partially Observable Markov Decision Processes (POMDPs) are a natural and general model in reinforcement learning that take into account the agent's uncertainty about its current s…