AI-Driven Detection of Industry-Specific Financial Fraud: A Theory-Informed NLP Approach
Combines topic modelling and supervised learning on SEC enforcement releases to classify five fraud types and map them to industry sectors.
Read the paperRoyal Roads University · School of Business
Building AI that finds the signal in complex data.
Dr. Mark Lokanan is a data scientist and machine learning researcher. His work spans predictive analytics, natural language processing, generative AI, and AI for business and financial crime.
data pointflagged anomaly
Research areas
Applied AI across four connected fields
Each area pairs rigorous modelling with real-world data from regulators, companies, and public institutions.
ML · 01Machine Learning & Predictive Analytics
Supervised and ensemble models that forecast outcomes and rank risk, with attention to data preparation and interpretability.
NLP · 02NLP & Generative AI
Topic modelling, text classification, and LLM applications that turn regulatory filings, testimony, and disclosures into evidence.
FCA · 03AI for Financial Crime
Fraud detection, anti-money laundering, and regulatory technology, built on two decades of enforcement and investigation research.
BIZ · 04AI in Business Functions
Analytics for the decisions organizations make every day, from hiring and marketing to supply chain integrity.
Responsible AI
How the models are built
Explainable
Every model has to show why it flags a case, so analysts and regulators can check the reasoning.
Fair
Models are tested for bias across groups before they inform decisions about people or firms.
Private and secure
Sensitive financial and video data is minimized, protected, and used only for the research purpose agreed.
Recent work
Featured publications
Supply Chain Fraud Prediction with Machine Learning and Artificial Intelligence
Tests ML and AI models for detecting irregularities in digitized supply chains. International Journal of Production Research, 63(1).
Read the paperThe Use of Machine Learning Algorithms to Predict Financial Statement Fraud
With Satish Sharma. Compares classifiers for flagging fraudulent financial statements. British Accounting Review.
Read the paperFull publication list · journal articles, book chapters, and working papers, updated automatically.
View all on Google ScholarProjects
Current research projects
Applied AI work in progress with industry and research partners.
FinGuard AI
Financial distress · Earnings manipulationMachine learning and explainable AI that flag companies at risk of financial distress and signs of earnings manipulation, built on peer-reviewed research.
Supported by Mitacs
Visit fin-guard.ai
Computer Vision for Theft Detection
Video analytics · Loss preventionDeep learning models that recognize theft behaviour in video footage, helping organizations detect and respond to losses.
Impact
Research that reaches policy and practice
Policy
Recommendations submitted to Ontario's Capital Markets Modernization Taskforce, drawn from enforcement research on Canada's securities regulators.
Industry
A fraud prediction model built on investment-industry enforcement data, developed into software for use by industry.
Public conversation
Commentary and media coverage on AI in fraud detection, money laundering in Canadian real estate, and investor protection.
Research to practice
Where the research goes to work
Findings from the lab move into enterprise AI solutions for organizations across sectors.
CEO
Vedia
Enterprise AI, analytics, and financial crime intelligence. Vedia builds AI agents and automation, predictive analytics, and fraud and AML systems for organizations across sectors.
- Vedia AI
- Vedia Analytics
- Vedia Secure
- Technology Services
About
Dr. Mark Lokanan
Mark Lokanan is a professor in the School of Business at Royal Roads University and a machine learning researcher whose work moves between theory, data, and deployment. He builds regression, classification, clustering, and language models, and applies them wherever better prediction changes decisions.
Before academia, he worked as a fraud investigator for Ontario's Ministry of the Attorney General. That field experience still shapes how he frames problems: start with the real process, then model it.
Degrees
- PhD, Fraud AnalyticsSimon Fraser University
- MA, Statistics and AnalyticsUniversity of Guelph
- MSc, Computer Science (Applied Data Science)Dublin
- MSc, Criminal Networks (Text Analytics)San Diego
- MSc, Forensic Accounting (Statistics and Data Modelling)Portsmouth
- BA, Sociology and LawYork University
Certificates
- Graduate Certificate in Deep Learning for AICarnegie Mellon University, School of Computer Science
- Post Graduate Certificate in Artificial IntelligenceUniversity of Texas
- Certificate in Artificial Intelligence ApplicationNorthwestern University
- Certificate in Machine LearningMIT
- Certificate in Generative AIMicrosoft
- Professional Certificate in Machine LearningIBM
- Certificate in Data ScienceEmeritus Institute of Management
- Certificate in Data AnalystSpringboard / Microsoft
- Advanced Certificate in AccountingMcMaster University
- Advanced Certificate in Forensic AccountingBCIT
Speaking and media
Expert commentary on applied AI
Available for interviews, keynotes, panels, and executive briefings on:
- AI and machine learning for fraud detection
- Generative AI and LLMs in business
- Anti-money laundering and financial crime analytics
- Computer vision for security and loss prevention
- Explainable and responsible AI
- Predicting corporate financial distress
Media and speaking requests: mark.lokanan@royalroads.ca
Partner with the lab
Co-fund applied AI research through Mitacs
Mitacs funds research partnerships between Canadian organizations and universities. Partners bring a real problem and data, and the lab builds and tests models with them. FinGuard AI is funded through a Mitacs grant.
mark.lokanan@royalroads.ca