Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

An Open API Architecture to Discover the Trustworthy Explanation of Cloud AI Services

Zerui Wang, Yan Liu, Jun Huang

TL;DR

This article presents the design of an open-API-based explainable AI (XAI) service to provide feature contribution explanations for cloud AI services without unfolding the network structure of the cloud models, and confirms that the architecture is cloud-agnostic.

Abstract

This article presents the design of an open-API-based explainable AI (XAI) service to provide feature contribution explanations for cloud AI services. Cloud AI services are widely used to develop domain-specific applications with precise learning metrics. However, the underlying cloud AI services remain opaque on how the model produces the prediction. We argue that XAI operations are accessible as open APIs to enable the consolidation of the XAI operations into the cloud AI services assessment. We propose a design using a microservice architecture that offers feature contribution explanations for cloud AI services without unfolding the network structure of the cloud models. We can also utilize this architecture to evaluate the model performance and XAI consistency metrics showing cloud AI services trustworthiness. We collect provenance data from operational pipelines to enable reproducibility within the XAI service. Furthermore, we present the discovery scenarios for the experimental tests regarding model performance and XAI consistency metrics for the leading cloud vision AI services. The results confirm that the architecture, based on open APIs, is cloud-agnostic. Additionally, data augmentations result in measurable improvements in XAI consistency metrics for cloud AI services.

An Open API Architecture to Discover the Trustworthy Explanation of Cloud AI Services

TL;DR

This article presents the design of an open-API-based explainable AI (XAI) service to provide feature contribution explanations for cloud AI services without unfolding the network structure of the cloud models, and confirms that the architecture is cloud-agnostic.

Abstract

This article presents the design of an open-API-based explainable AI (XAI) service to provide feature contribution explanations for cloud AI services. Cloud AI services are widely used to develop domain-specific applications with precise learning metrics. However, the underlying cloud AI services remain opaque on how the model produces the prediction. We argue that XAI operations are accessible as open APIs to enable the consolidation of the XAI operations into the cloud AI services assessment. We propose a design using a microservice architecture that offers feature contribution explanations for cloud AI services without unfolding the network structure of the cloud models. We can also utilize this architecture to evaluate the model performance and XAI consistency metrics showing cloud AI services trustworthiness. We collect provenance data from operational pipelines to enable reproducibility within the XAI service. Furthermore, we present the discovery scenarios for the experimental tests regarding model performance and XAI consistency metrics for the leading cloud vision AI services. The results confirm that the architecture, based on open APIs, is cloud-agnostic. Additionally, data augmentations result in measurable improvements in XAI consistency metrics for cloud AI services.

Paper Structure

This paper contains 36 sections, 2 equations, 9 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Related Works
  3. Cloud-based AI Services
  4. The Post-hoc XAI Methods
  5. Frameworks and Packages for Implementing XAI
  6. Data Provenance for XAI
  7. Background
  8. CAM-based XAI methods
  9. Metrics for Evaluating XAI Techniques
  10. The Prediction Change Value
  11. XAI Consistency Evaluation
  12. The Post hoc XAI Methods with Feature Contribution Explanation
  13. XAI Methods Taxonomy
  14. Integrating Cloud AI with Post hoc XAI
  15. XAI Service Architecture and API Design
  16. ...and 21 more sections

Figures (9)

  • Figure 1: Grad-CAM Visual Explanations on ImageNet Dataset Samples
  • Figure 2: Workflow for Integrating Cloud AI Services with State-of-the-Art Deep Learning Models to Approximate Feature Contribution Values Using Post-Hoc XAI Methods
  • Figure 3: Microservice-Based Reference Architecture for XAI Service
  • Figure 4: Directed Graph Model of XAI Provenance Data Capturing Microservices, Tasks, and Pipeline Relationships
  • Figure 5: Comparison of Prediction Change Values for ResNet and DenseNet Approximation Models in CAM-based XAI Methods
  • ...and 4 more figures