Functional Transparency for Structured Data: a Game-Theoretic Approach

02/26/2019
by   Guang-He Lee, et al.
0

We provide a new approach to training neural models to exhibit transparency in a well-defined, functional manner. Our approach naturally operates over structured data and tailors the predictor, functionally, towards a chosen family of (local) witnesses. The estimation problem is setup as a co-operative game between an unrestricted predictor such as a neural network, and a set of witnesses chosen from the desired transparent family. The goal of the witnesses is to highlight, locally, how well the predictor conforms to the chosen family of functions, while the predictor is trained to minimize the highlighted discrepancy. We emphasize that the predictor remains globally powerful as it is only encouraged to agree locally with locally adapted witnesses. We analyze the effect of the proposed approach, provide example formulations in the context of deep graph and sequence models, and empirically illustrate the idea in chemical property prediction, temporal modeling, and molecule representation learning.

READ FULL TEXT

page 1

page 2

page 3

page 4

research
06/30/2018

Game-Theoretic Interpretability for Temporal Modeling

Interpretability has arisen as a key desideratum of machine learning mod...
research
02/04/2022

Functional Mixtures-of-Experts

We consider the statistical analysis of heterogeneous data for clusterin...
research
03/26/2023

Prediction and estimation of random variables with infinite mean or variance

In this paper we propose an optimal predictor of a random variable that ...
research
02/05/2018

Proposed Spreadsheet Transparency Definition and Measures

Auditors demand financial models be transparent yet no consensus exists ...
research
09/20/2020

Latent Representation Prediction Networks

Deeply-learned planning methods are often based on learning representati...
research
02/21/2017

Best Linear Predictor with Missing Response: Locally Robust Approach

This paper provides asymptotic theory for Inverse Probability Weighing (...
research
11/04/2021

Testing using Privileged Information by Adapting Features with Statistical Dependence

Given an imperfect predictor, we exploit additional features at test tim...

Please sign up or login with your details

Forgot password? Click here to reset