Pie & AI Suisse - Trustworthiness of AI models: Improving NLP with Causality

Abstract:
This session focuses on how causality can be used to improve NLP models like BERT, and especially some traditional NLP tasks such as text classification, sentiment analysis, or text generation.

Papers & articles:

• Causal Inference in Natural Language Processing: Estimation, Prediction, Interpretation and Beyond https://arxiv.org/abs/2109.00725
• Introducing BERT, ELMo, & Co. https://jalammar.github.io/illustrated-bert/
• Formalizing Trust in Artificial Intelligence: Prerequisites, Causes and Goals of Human Trust in AI https://arxiv.org/abs/2010.07487

Speaker:
Haaya Naushan is a data scientist focused on integrating data science and machine learning with socially conscious research. She currently works as a World Bank data science consultant, where she uses NLP, ML and econometrics for economics research. Her experience includes leveraging NLP and graph algorithms with big data, to study social media around the topics of disinformation and hate speech. Haaya is keenly interested in AI Ethics, and often incorporates aspects of social justice into her data science articles on Medium. She is also fascinated by causal inference, so she is always looking for innovative ways to use causality in her research.

+++ Code of Conduct ++++++++++++
https://github.com/microsoft/virtual-events/blob/main/virtual-event-code-of-conduct.md

+++ AI Suisse YouTube Channel ++++
Sign up for our YouTube Channel: https://www.youtube.com/c/AISuisse

+++ deeplearning.ai ++++++++++++++
This is a Pie & AI event in partnership with DeepLearning.AI. Pie & AI is a series of DeepLearning.AI meetups independently hosted by community groups. This event is hosted by AI Suisse. In addition to signing up on Meetup.com, sign up here (https://www.eventbrite.com/e/191182661487) if you'd like to receive a course promo code and future event updates from DeepLearning.AI