基本信息
浏览量:8
职业迁徙
个人简介
My main focus is to make NLP models more robust. For instance, I have worked on analysis
and debiasing algorithms for gender biases in BERT where we found directions in the embedding space that encode gender information. Another work comes from the interpretability of Transformer where we theoretically analyse if the self-attention is identifiable and
propose identifiable transformers (accepted at ACL 2021). My under review submission to
a reputed conference includes Optimal Transport (OT) based algorithms that confidently
distill unbiased knowledge from an intelligent system to another. I was fortunate to work
at Salesforce as NLP research intern where I worked on parameter-efficient prompt-tuning.
I am currently visiting California, working as an Applied Scientist Intern at Amazon, California, USA.
and debiasing algorithms for gender biases in BERT where we found directions in the embedding space that encode gender information. Another work comes from the interpretability of Transformer where we theoretically analyse if the self-attention is identifiable and
propose identifiable transformers (accepted at ACL 2021). My under review submission to
a reputed conference includes Optimal Transport (OT) based algorithms that confidently
distill unbiased knowledge from an intelligent system to another. I was fortunate to work
at Salesforce as NLP research intern where I worked on parameter-efficient prompt-tuning.
I am currently visiting California, working as an Applied Scientist Intern at Amazon, California, USA.
研究兴趣
论文共 36 篇作者统计合作学者相似作者
按年份排序按引用量排序主题筛选期刊级别筛选合作者筛选合作机构筛选
时间
引用量
主题
期刊级别
合作者
合作机构
CoRR (2024)
引用0浏览0EI引用
0
0
Kritika Johari, Christopher Tay Zi Tong,Rishabh Bhardwaj,Vigneshwaran Subbaraju,Jung-Jae Kim,U.-Xuan Tan
The Visual Computerpp.1-15, (2024)
arXiv (Cornell University) (2023)
引用0浏览0引用
0
0
EMNLP 2023 (2023): 13546-13557
引用0浏览0EI引用
0
0
加载更多
作者统计
合作学者
合作机构
D-Core
- 合作者
- 学生
- 导师
数据免责声明
页面数据均来自互联网公开来源、合作出版商和通过AI技术自动分析结果,我们不对页面数据的有效性、准确性、正确性、可靠性、完整性和及时性做出任何承诺和保证。若有疑问,可以通过电子邮件方式联系我们:report@aminer.cn