The Mythos of Model Interpretability

نویسنده

Zachary Chase Lipton

چکیده

Supervised machine learning models boast remarkable predictive capabilities. But can you trust your model? Will it work in deployment? What else can it tell you about the world? We want models to be not only good, but interpretable. And yet the task of interpretation appears underspecified. Papers provide diverse and sometimes non-overlapping motivations for interpretability, and offer myriad notions of what attributes render models interpretable. Despite this ambiguity, many papers proclaim interpretability axiomatically, absent further explanation. In this paper, we seek to refine the discourse on interpretability. First, we examine the motivations underlying interest in interpretability, finding them to be diverse and occasionally discordant. Then, we address model properties and techniques thought to confer interpretability, identifying transparency to humans and post-hoc explanations as competing notions. Throughout, we discuss the feasibility and desirability of different notions, and question the oft-made assertions that linear models are interpretable and that deep neural networks are not.

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

SECURING INTERPRETABILITY OF FUZZY MODELS FOR MODELING NONLINEAR MIMO SYSTEMS USING A HYBRID OF EVOLUTIONARY ALGORITHMS

In this study, a Multi-Objective Genetic Algorithm (MOGA) is utilized to extract interpretable and compact fuzzy rule bases for modeling nonlinear Multi-input Multi-output (MIMO) systems. In the process of non- linear system identi cation, structure selection, parameter estimation, model performance and model validation are important objectives. Furthermore, se- curing low-level and high-level ...

متن کامل

A Rule Extractor for Diagnosing the Type 2 Diabetes Using a Self-organizing Genetic Algorithm

Introduction: Constructing medical decision support models to automatically extract knowledge from data helps physicians in early diagnosis of disease. Interpretability of the inferential rules of these models is a key indicator in determining their performance in order to understand how they make decisions, and increase the reliability of their output. Methods: In this study, an automated hyb...

متن کامل

Tying Victorian Bond(s) for the Resurrection of Her Majesty’s Secret Service

The James Bond phenomenon has been approached to discuss notions of British identity, imperialistic ethos, issues of masculinity and commodified middle-class. However, the critical canon of the James Bond mythos has not delved into the ideological origins of the myth, namely Thomas Carlyle’s (1795-1881) myth of the Hero, the “blueprint” of Victorian ideology. The article explores Sam Mendes’ Sk...

متن کامل