A Formal Approach to Explainability

01/15/2020
by   Lior Wolf, et al.
0

We regard explanations as a blending of the input sample and the model's output and offer a few definitions that capture various desired properties of the function that generates these explanations. We study the links between these properties and between explanation-generating functions and intermediate representations of learned models and are able to show, for example, that if the activations of a given layer are consistent with an explanation, then so do all other subsequent layers. In addition, we study the intersection and union of explanations as a way to construct new explanations.

READ FULL TEXT

page 1

page 2

page 3

page 4

research
12/29/2022

A Theoretical Framework for AI Models Explainability

Explainability is a vibrant research topic in the artificial intelligenc...
research
04/28/2023

Interpreting Vision and Language Generative Models with Semantic Visual Priors

When applied to Image-to-text models, interpretability methods often pro...
research
04/25/2022

Generating and Visualizing Trace Link Explanations

Recent breakthroughs in deep-learning (DL) approaches have resulted in t...
research
06/22/2020

Fanoos: Multi-Resolution, Multi-Strength, Interactive Explanations for Learned Systems

Machine learning becomes increasingly important to tune or even synthesi...
research
12/10/2020

Influence-Driven Explanations for Bayesian Network Classifiers

One of the most pressing issues in AI in recent years has been the need ...
research
11/01/2019

What Gets Echoed? Understanding the "Pointers" in Explanations of Persuasive Arguments

Explanations are central to everyday life, and are a topic of growing in...
research
05/29/2017

Contextual Explanation Networks

We introduce contextual explanation networks (CENs)---a class of models ...

Please sign up or login with your details

Forgot password? Click here to reset