Welcome to my homepage! I keep a collection of my research projects here.
I am interested in building robust and reliable AI systems. I am currently working on adversarial robustness as an AI Resident at MSR AI. I worked on how bet...
- 2020 2
- 2019 2
- 2018 2
2020
Randomized Smoothing Done Right
This is one of my first projects during my AI Residency at Microsoft Research.
2019
ParaBank 2: pushing the limit of diversity with sampling and clustering
Should paraphrasing use the same generation pipeline as machine translation? A question whose answer has been taken for granted. We suggest that paraphrasing...
Faster constrained decoding and better sentence rewriting
Constrained decoding, which we used in ParaBank, is not quite fast enough to be useful. We came up with an algorithmic improvement to boost its performance b...
2018
ParaBank: paraphrase generation from bilingual data
Paraphrases are quite useful for understanding semantics, yet most resources are lexical. We combined back-translation (from ParaNMT) and constrained decodin...
DNC: probling NLI models with recasted datasets
My first NLP project. We recasted a lot of NLP datasets into the NLI format - premise and hypothesis, and used them to probe what kind of inferences NLI mode...