Attention Mechanisms Don't Learn Additive Models: Rethinking Feature Importance for Transformers

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Leemann, Tobias, Fastowski, Alina, Pfeiffer, Felix, Kasneci, Gjergji
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912180727709696
author Leemann, Tobias
Fastowski, Alina
Pfeiffer, Felix
Kasneci, Gjergji
author_facet Leemann, Tobias
Fastowski, Alina
Pfeiffer, Felix
Kasneci, Gjergji
contents We address the critical challenge of applying feature attribution methods to the transformer architecture, which dominates current applications in natural language processing and beyond. Traditional attribution methods to explainable AI (XAI) explicitly or implicitly rely on linear or additive surrogate models to quantify the impact of input features on a model's output. In this work, we formally prove an alarming incompatibility: transformers are structurally incapable of representing linear or additive surrogate models used for feature attribution, undermining the grounding of these conventional explanation methodologies. To address this discrepancy, we introduce the Softmax-Linked Additive Log Odds Model (SLALOM), a novel surrogate model specifically designed to align with the transformer framework. SLALOM demonstrates the capacity to deliver a range of insightful explanations with both synthetic and real-world datasets. We highlight SLALOM's unique efficiency-quality curve by showing that SLALOM can produce explanations with substantially higher fidelity than competing surrogate models or provide explanations of comparable quality at a fraction of their computational costs. We release code for SLALOM as an open-source project online at https://github.com/tleemann/slalom_explanations.
format Preprint
id arxiv_https___arxiv_org_abs_2405_13536
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Attention Mechanisms Don't Learn Additive Models: Rethinking Feature Importance for Transformers
Leemann, Tobias
Fastowski, Alina
Pfeiffer, Felix
Kasneci, Gjergji
Machine Learning
Artificial Intelligence
Computation and Language
We address the critical challenge of applying feature attribution methods to the transformer architecture, which dominates current applications in natural language processing and beyond. Traditional attribution methods to explainable AI (XAI) explicitly or implicitly rely on linear or additive surrogate models to quantify the impact of input features on a model's output. In this work, we formally prove an alarming incompatibility: transformers are structurally incapable of representing linear or additive surrogate models used for feature attribution, undermining the grounding of these conventional explanation methodologies. To address this discrepancy, we introduce the Softmax-Linked Additive Log Odds Model (SLALOM), a novel surrogate model specifically designed to align with the transformer framework. SLALOM demonstrates the capacity to deliver a range of insightful explanations with both synthetic and real-world datasets. We highlight SLALOM's unique efficiency-quality curve by showing that SLALOM can produce explanations with substantially higher fidelity than competing surrogate models or provide explanations of comparable quality at a fraction of their computational costs. We release code for SLALOM as an open-source project online at https://github.com/tleemann/slalom_explanations.
title Attention Mechanisms Don't Learn Additive Models: Rethinking Feature Importance for Transformers
topic Machine Learning
Artificial Intelligence
Computation and Language
url https://arxiv.org/abs/2405.13536