Bridging Accuracy and Accountability: Interpretable AI for Fair and Transparent Crime Analytics

Read full text Summarise with VIRAT
Authors: B. Yamini Priyanka, Oleti Deepika, Gokavarapu Harika, Ponku Vardhan Babu, Shaik Muskan Bhanu
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Artificial Intelligence (AI) has revolutionized crime prediction, enabling law enforcement to analyze vast datasets for patterns in crimes against persons, property, and society, primarily using public datasets from U.S. cities like Chicago. Existing systems employ traditional Machine Learning (ML) techniques such as Random Forest (42% usage), Deep Learning (DL) models like LSTM (26%), GIS hotspot analysis (20%), and statistical methods (13%), achieving high accuracy in spatio-temporal hotspot prediction (36.62% of studies). However, these systems face limitations including insufficient explainability in opaque DL models, limited dataset availability lacking time/location details, bias perpetuation affecting disadvantaged groups, privacy risks, and underdeveloped XAI integration, hindering trust and ethical deployment.The proposed system addresses these gaps through advanced Explainable AI (XAI) techniques like SHAP and LIME (post-hoc), alongside interpretable-by-design models (ante-hoc), hybrid ML-DL-graph neural networks (e.g., STDGCN, AIST), federated learning for privacy, and diverse high-quality datasets. Benefits include enhanced transparency for decision accountability, bias mitigation for fairness, improved accuracy via interdisciplinary benchmarks, real-time geospatial visualization, and ethical AI fostering public trust and equitable policing.

Fields of Study & Subjects

Explainable Artificial Intelligence (XAI) Ethics and Social Impacts of AI Artificial Intelligence in Law

Publication details

Journal / source
Zenodo (CERN European Organization for Nuclear Research)
Publication date
22 March 2026
Language
EN