Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Nishpaksh: TEC Standard-Compliant Framework for Fairness Auditing and Certification of AI Models

Shashank Prakash, Ranjitha Prasad, Avinash Agarwal

TL;DR

This work addresses the need for region-specific, auditable fairness assessment in AI systems deployed in telecom and 6G by introducing Nishpaksh, a TEC-standard compliant fairness auditing tool. It unifies risk quantification, contextual thresholding, and bias testing into a web-based dashboard that outputs audit-grade reports and certification-ready scores. The framework operationalizes the TEC standard through a seven-domain lifecycle risk approach, a five-point risk scale, and BI/FS metrics, validated on the COMPAS dataset to reveal attribute-specific bias and demonstrate reproducible assessments. Nishpaksh bridges global fairness methodologies with regulatory AI governance in India, offering an indigenous solution for self-certification and third-party auditing and outlining a path toward broader standardization in telecom AI governance.

Abstract

The growing reliance on Artificial Intelligence (AI) models in high-stakes decision-making systems, particularly within emerging telecom and 6G applications, underscores the urgent need for transparent and standardized fairness assessment frameworks. While global toolkits such as IBM AI Fairness 360 and Microsoft Fairlearn have advanced bias detection, they often lack alignment with region-specific regulatory requirements and national priorities. To address this gap, we propose Nishpaksh, an indigenous fairness evaluation tool that operationalizes the Telecommunication Engineering Centre (TEC) Standard for the Evaluation and Rating of Artificial Intelligence Systems. Nishpaksh integrates survey-based risk quantification, contextual threshold determination, and quantitative fairness evaluation into a unified, web-based dashboard. The tool employs vectorized computation, reactive state management, and certification-ready reporting to enable reproducible, audit-grade assessments, thereby addressing a critical post-standardization implementation need. Experimental validation on the COMPAS dataset demonstrates Nishpaksh's effectiveness in identifying attribute-specific bias and generating standardized fairness scores compliant with the TEC framework. The system bridges the gap between research-oriented fairness methodologies and regulatory AI governance in India, marking a significant step toward responsible and auditable AI deployment within critical infrastructure like telecommunications.

Nishpaksh: TEC Standard-Compliant Framework for Fairness Auditing and Certification of AI Models

TL;DR

This work addresses the need for region-specific, auditable fairness assessment in AI systems deployed in telecom and 6G by introducing Nishpaksh, a TEC-standard compliant fairness auditing tool. It unifies risk quantification, contextual thresholding, and bias testing into a web-based dashboard that outputs audit-grade reports and certification-ready scores. The framework operationalizes the TEC standard through a seven-domain lifecycle risk approach, a five-point risk scale, and BI/FS metrics, validated on the COMPAS dataset to reveal attribute-specific bias and demonstrate reproducible assessments. Nishpaksh bridges global fairness methodologies with regulatory AI governance in India, offering an indigenous solution for self-certification and third-party auditing and outlining a path toward broader standardization in telecom AI governance.

Abstract

The growing reliance on Artificial Intelligence (AI) models in high-stakes decision-making systems, particularly within emerging telecom and 6G applications, underscores the urgent need for transparent and standardized fairness assessment frameworks. While global toolkits such as IBM AI Fairness 360 and Microsoft Fairlearn have advanced bias detection, they often lack alignment with region-specific regulatory requirements and national priorities. To address this gap, we propose Nishpaksh, an indigenous fairness evaluation tool that operationalizes the Telecommunication Engineering Centre (TEC) Standard for the Evaluation and Rating of Artificial Intelligence Systems. Nishpaksh integrates survey-based risk quantification, contextual threshold determination, and quantitative fairness evaluation into a unified, web-based dashboard. The tool employs vectorized computation, reactive state management, and certification-ready reporting to enable reproducible, audit-grade assessments, thereby addressing a critical post-standardization implementation need. Experimental validation on the COMPAS dataset demonstrates Nishpaksh's effectiveness in identifying attribute-specific bias and generating standardized fairness scores compliant with the TEC framework. The system bridges the gap between research-oriented fairness methodologies and regulatory AI governance in India, marking a significant step toward responsible and auditable AI deployment within critical infrastructure like telecommunications.
Paper Structure (14 sections, 7 equations, 5 figures, 2 tables)

This paper contains 14 sections, 7 equations, 5 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Related Works
  3. TEC Standard and Evaluation
  4. Lifecycle based assessment and risk quantification
  5. Contextual Threshold Determination
  6. Inference and Results Architecture
  7. Design and Architecture of Nishpaksh
  8. Experiments and Results
  9. Dataset Description
  10. Experimental Setup
  11. Fairness Metrics and Formulations
  12. Fairness Metrics Results and Visualizations
  13. Implications for Standardization
  14. Conclusion

Figures (5)

  • Figure 1: User interface of the Nishpaksh showing the survey intake and configuration panels.
  • Figure 2: Inference and Fairness Evaluation interface of the Nishpaksh showing model wise disparity analysis for sensitive attribute.
  • Figure 3: Distribution of sensitive attributes in the COMPAS dataset.
  • Figure 4: Fairness metric comparison across bias exposure conditions.
  • Figure 5: Subgroup misclassification rates by race and gender.