The disagreement issue in post hoc feature attribution techniques is discussed in this study. Explainers like SHAP, LIME, and gradient-based techniques frequently result in contradictory feature importance rankings for the same model. Post hoc Explainer Agreement Regularization (PEAR), a loss term added after model training, is introduced to counteract this and promote increased explainer consensus without significantly compromising accuracy. Experiments on three datasets show that PEAR offers a customizable balance between explanation consensus and predictive performance, and it enhances agreement across explainers, including those not directly used in training. PEAR improves explanations' dependability and credibility in crucial machine learning applications by turning disagreement into a controlled parameter.The disagreement issue in post hoc feature attribution techniques is discussed in this study. Explainers like SHAP, LIME, and gradient-based techniques frequently result in contradictory feature importance rankings for the same model. Post hoc Explainer Agreement Regularization (PEAR), a loss term added after model training, is introduced to counteract this and promote increased explainer consensus without significantly compromising accuracy. Experiments on three datasets show that PEAR offers a customizable balance between explanation consensus and predictive performance, and it enhances agreement across explainers, including those not directly used in training. PEAR improves explanations' dependability and credibility in crucial machine learning applications by turning disagreement into a controlled parameter.

New AI Study Tackles the Transparency Problem in Black-Box Models

2025/09/21 13:46

:::info Authors:

(1) Avi Schwarzschild, University of Maryland, College Park, Maryland, USA and Work completed while working at Arthur (avi1umd.edu);

(2) Max Cembalest, Arthur, New York City, New York, USA;

(3) Karthik Rao, Arthur, New York City, New York, USA;

(4) Keegan Hines, Arthur, New York City, New York, USA;

(5) John Dickerson†, Arthur, New York City, New York, USA ([email protected]).

:::

Abstract and 1. Introduction

1.1 Post Hoc Explanation

1.2 The Disagreement Problem

1.3 Encouraging Explanation Consensus

  1. Related Work

  2. Pear: Post HOC Explainer Agreement Regularizer

  3. The Efficacy of Consensus Training

    4.1 Agreement Metrics

    4.2 Improving Consensus Metrics

    [4.3 Consistency At What Cost?]()

    4.4 Are the Explanations Still Valuable?

    4.5 Consensus and Linearity

    4.6 Two Loss Terms

  4. Discussion

    5.1 Future Work

    5.2 Conclusion, Acknowledgements, and References

Appendix

ABSTRACT

As neural networks increasingly make critical decisions in highstakes settings, monitoring and explaining their behavior in an understandable and trustworthy manner is a necessity. One commonly used type of explainer is post hoc feature attribution, a family of methods for giving each feature in an input a score corresponding to its influence on a model’s output. A major limitation of this family of explainers in practice is that they can disagree on which features are more important than others. Our contribution in this paper is a method of training models with this disagreement problem in mind. We do this by introducing a Post hoc Explainer Agreement Regularization (PEAR) loss term alongside the standard term corresponding to accuracy, an additional term that measures the difference in feature attribution between a pair of explainers. We observe on three datasets that we can train a model with this loss term to improve explanation consensus on unseen data, and see improved consensus between explainers other than those used in the loss term. We examine the trade-off between improved consensus and model performance. And finally, we study the influence our method has on feature attribution explanations.

1 INTRODUCTION

As machine learning becomes inseparable from important societal sectors like healthcare and finance, increased transparency of how complex models arrive at their decisions is becoming critical. In this work, we examine a common task in support of model transparency that arises with the deployment of complex black-box models in production settings: explaining which features in the input are most influential in the model’s output. This practice allows data scientists and machine learning practitioners to rank features by importance – the features with high impact on model output are considered more important, and those with little impact on model output are considered less important. These measurements inform how model users debug and quality check their models, as well as how they explain model behavior to stakeholders.

1.1 Post Hoc Explanation

The methods of model explanation considered in this paper are post hoc local feature attribution scores. The field of explainable artificial intelligence (XAI) is rapidly producing different methods of this

\ Figure 1: Our loss that encourages explainer consensus boosts the correlation between LIME and other common post hoc explainers. This comes with a cost of less than two percentage points of accuracy compared with our baseline model on the Electricity dataset. Our method improves consensus on six agreement metrics and all pairs of explainers we evaluated. Note that this plot measures the rank correlation agreement metric and the specific bar heights depend on this choice of metric.

\ type to make sense of model behavior [e.g., 21, 24, 30, 32, 37]. Each of these methods has a slightly different formula and interpretation of its raw output, but in general they all perform the same task of attributing a model’s behavior to its input features. When tasked to explain a model’s output with a corresponding input (and possible access to the model weights), these methods answer the question, “How influential is each individual feature of the input in the model’s computation of the output?”

\ Data scientists are using post hoc explainers at increasing rates – popular methods like LIME and SHAP have had over 350 thousand and 6 million downloads of their Python packages in the last 30 days, respectively [23].

1.2 The Disagreement Problem

The explosion of different explanation methods leads Krishna et al. [15] to observe that when neural networks are trained naturally, i.e. for accuracy alone, often post hoc explainers disagree on how much different features influenced a model’s outputs. They coin the term the disagreement problem and argue that when explainers disagree about which features of the input are important, practitioners have little concrete evidence as to which of the explanations, if any, to trust.

\ There is an important discussion around local explainers and their true value in reaching the communal goal of model transparency and interpretability [see, e.g., 7, 18, 29]; indeed, there are ongoing discussions about the efficacy of present-day explanation methods in specific domains [for healthcare see, e.g., 8]. Feature importance estimates may fail at making a model more transparent when the model being explained is too complex to allow for easily attributing the output to the contribution of each individual feature.

\ In this paper, we make no normative judgments with respect to this debate, but rather view “explanations” as signals to be used alongside other debugging, validation, and verification approaches in the machine learning operations (MLOps) pipeline. Specifically, we take the following practical approach: make the amount of explanation disagreement a controllable model parameter instead of a point of frustration that catches stakeholders off-guard.

1.3 Encouraging Explanation Consensus

Consensus between two explainers does not require that the explainers output the same exact scores for each feature. Rather, consensus between explainers means that whatever disagreement they exhibit can be reconciled. Data scientists and machine learning practitioners say in a survey that explanations are in basic agreement if they satisfy agreement metrics that align with human intuition, which provides a quantitative way to evaluate the extent to which consensus is being achieved [15]. When faced with disagreement between explainers, a choice has to be made about what to do next – if such an arbitrary crossroads moment is avoidable via specialized model training, we believe it would be a valuable addition to a data scientist’s toolkit.

\ We propose, as our main contribution, a training routine to help alleviate the challenge posed by post hoc explanation disagreement. Achieving better consensus between explanations does not provide more interpretability to a model inherently. But, it may lend more trust to the explanations if different approaches to attribution agree more often on which features are important. This gives consensus the practical benefit of acting as a sanity check – if consensus is observed, the choice of which explainer a practitioner uses is less consequential with respect to downstream stakeholder impact, making their interpretation less subjective.

2 RELATED WORK

Our work focuses on post hoc explanation tools. Some post hoc explainers, like LIME [24] and SHAP [21], are proxy models trained atop a base machine learning model with the sole intention of “explaining” that base model. These explainers rely only on the model’s inputs and outputs to identify salient features. Other explainers, such as Vanilla Gradients (Grad) [32], Gradient Times Input (Grad*Input) [30], Integrated Gradients (IntGrad) [37] and SmoothGrad [34], do not use a proxy model but instead compute the gradients of a model with respect to input features to identify important features.[1] Each of these explainers has its quirks and there are reasons to use, or not use, them all—based on input type, model type, downstream task, and so on. But there is an underlying pattern unifying all these explanation tools. Han et al. [12] provide a framework that characterizes all the post hoc explainers used in this paper as different types of local-function approximation. For more details about the individual post hoc explainers used in this paper, we refer the reader to the individual papers and to other works about when and why to use each one [see, e.g., 5, 13].

\ We build directly on prior work that defines and explores the disagreement problem [15]. Disagreement here refers to the difference in feature importance scores between two feature attribution methods, but can be quantified several different ways as are described by the metrics Krishna et al. [15] define and use. We describe these metrics in Section 4.

\ The method we propose in this paper relates to previous work that trains models with constraints on explanations via penalties on the disagreement between feature attribution scores and handcrafted ground-truth scores [26, 27, 41]. Additionally, work has been done to leverage the disagreement between different posthoc explanations to construct new feature attribution scores that improve metrics like stability and pairwise rank agreement [2, 16, 25].

\

:::info This paper is available on arxiv under CC BY 4.0 DEED license.

:::

\

Disclaimer: The articles reposted on this site are sourced from public platforms and are provided for informational purposes only. They do not necessarily reflect the views of MEXC. All rights remain with the original authors. If you believe any content infringes on third-party rights, please contact [email protected] for removal. MEXC makes no guarantees regarding the accuracy, completeness, or timeliness of the content and is not responsible for any actions taken based on the information provided. The content does not constitute financial, legal, or other professional advice, nor should it be considered a recommendation or endorsement by MEXC.
Share Insights

You May Also Like

Wormhole launches reserve tying protocol revenue to token

Wormhole launches reserve tying protocol revenue to token

The post Wormhole launches reserve tying protocol revenue to token appeared on BitcoinEthereumNews.com. Wormhole is changing how its W token works by creating a new reserve designed to hold value for the long term. Announced on Wednesday, the Wormhole Reserve will collect onchain and offchain revenues and other value generated across the protocol and its applications (including Portal) and accumulate them into W, locking the tokens within the reserve. The reserve is part of a broader update called W 2.0. Other changes include a 4% targeted base yield for tokenholders who stake and take part in governance. While staking rewards will vary, Wormhole said active users of ecosystem apps can earn boosted yields through features like Portal Earn. The team stressed that no new tokens are being minted; rewards come from existing supply and protocol revenues, keeping the cap fixed at 10 billion. Wormhole is also overhauling its token release schedule. Instead of releasing large amounts of W at once under the old “cliff” model, the network will shift to steady, bi-weekly unlocks starting October 3, 2025. The aim is to avoid sharp periods of selling pressure and create a more predictable environment for investors. Lockups for some groups, including validators and investors, will extend an additional six months, until October 2028. Core contributor tokens remain under longer contractual time locks. Wormhole launched in 2020 as a cross-chain bridge and now connects more than 40 blockchains. The W token powers governance and staking, with a capped supply of 10 billion. By redirecting fees and revenues into the new reserve, Wormhole is betting that its token can maintain value as demand for moving assets and data between chains grows. This is a developing story. This article was generated with the assistance of AI and reviewed by editor Jeffrey Albus before publication. Get the news in your inbox. Explore Blockworks newsletters: Source: https://blockworks.co/news/wormhole-launches-reserve
Share
2025/09/18 01:55