Big DataNights & Microsoft Reactor - Explainable AI #1: Introduction to XAI

Explainable AI (XAI) Course #1: Introduction to XAI
5 March, 2023 | 6:00 PM

Speakers: Maya Bercovitch and Rada Menczel
Speakers Language: Hebrew
Address: Microsoft Reactor Tel Aviv, Derech Menachem Begin 144, 'Midtown' Floor 50.

In this introduction lecture on explainability in AI, we will delve into the key topics that surround this emerging field. We will first provide an overview of the motivation for explainability, exploring how it helps us to achieve more transparent and trustworthy AI systems, particularly from a managerial perspective. We will then define some of the key terminology in the field and differentiate between black box explanation and interpretable ML. We will discuss the differences between global and local explanations and include many examples from different fields and use cases throughout the lecture. Next, we will examine the "built-in" feature of importance methods that are commonly used for regression and trees and discuss the strengths and limitations of these methods.

Overall, this lecture will provide a comprehensive introduction to explainability in AI, covering the key topics and terminology that are essential for understanding this field.

This session is part of a series of sessions in the XAI course.
This course is designed for data scientists that have at least two years of hands-on experience with machine learning and Python and a basic background in deep learning. The XAI course is managed on a voluntary basis by DataNights and Microsoft organizers and free for charge for the participant.
Please refer to the course sequel below:

• 13/03 | 19:00-20:30 - Local Explanations - Concept and Methods (Livestream, Hebrew)
• 19/03 | 18:00-21:00 - Explaining Deep Neural Network Models (In-person (Tel Aviv), Hebrew).
• 23/03 | 18:00-19:30 - Counterfactual Explanations - Explaining and Debugging (Livestream, English).
• 28/03 | 18:00-21:00 - Explainable AI in practice (In-person (Tel Aviv), Hebrew).

Course Leaders:
Bitya Neuhof, DataNights
Yasmin Bokobza, Microsoft