About

Kannan Srinivasan is the H.J. Heinz II Professor of Management, Marketing and Business Technologies at the Tepper School of Business. His role involves research and teaching in the areas of management, marketing, and business technologies, contributing to the school's strategic focus on integrating business, technology, and analytics. As a faculty member, he is engaged in advancing thought leadership and innovation within these fields, supporting the Tepper School's vision to lead at the intersection of business and technology.

Research topics

• Computer Science
• Machine Learning
• Economics
• Algorithm
• Microeconomics
• Business
• Sociology
• Computer Security
• Artificial Intelligence
• Political Science
• Advertising
• Geography
• Finance
• Marketing
• Psychology
• Social psychology
• Law

Selected publications

• An innovative framework for brain tumor detection and types classification using ScLe2 LU–DNN and GaPL-MOA

  Soft Computing · 2026-02-17

  article
  PublisherDOI

• A Robust XACT ² NN and ZGFR-Based Cultural Multi Linguistic-Aware Higher Educational Course Recommendation System

  Cybernetics & Systems · 2026-02-20

  article
  PublisherDOI

• Personalization, Consumer Search, and Algorithmic Pricing

  Marketing Science · 2025-05-07 · 6 citations

  articleSenior author

  This paper shows that personalized product rankings, although improving search relevance, can unintentionally enable AI pricing algorithms to raise prices and reduce consumer welfare.

  PublisherDOI

• Transforming Urban Landscapes with AI: Utilizing Reforestation Drones, Ocean Cleanup Robotics, Predictive Climate Modeling, and Green Infrastructure to Build Resilient and Sustainable Cities of Tomorrow

  Science for sustainable societies · 2025-01-01

  book-chapter
  PublisherDOI

• Innovative Cloud-Based E-Commerce Fraud Prevention Using GAN-FS, Fuzzy-Rough Clustering, Smart Contracts, and Game-Theoretic Models

  2025-07-04

  article

  The rise in online transactions will lead to increased fraud, often flagging legitimate users. Cloud systems provide extensive monitoring of large networks. By game theory, group analysis, GAN-FS synthetic generation, and leveraging blockchain verification, an adaptive intelligence protection system is created to respond to evolving threats. GAN-FS generates artificial bidding schemes as fuzzy-rough clustering detects subtle patterns in conflicting data. Game theory strengthens safeguards, while blockchain's self-enforcing contracts automate security, ensuring reliable cloud coverage and surveillance. This system enhances scalability, volumes, and real-time monitoring. Smart contracts secure transactions, fuzzy-rough clustering improves pattern recognition, and GANs with synthetic data improve fraud detection. It predicts fraudulent strategies using climate-scalable cloud architecture, secure, offering flexible, and game-theoretical simulators. In addition to automating transactions, synthetic data generation drives fraud detection. The proposed method achieved 91.8% scalability, 93.7% recall of real cases, and 94.3% classification accuracy. By combining game theory, fuzzy-rough clustering, GAN-FS, and blockchain smart contracts, false positives are reduced and real-time flexibility is enhanced. Hybrid frameworks are therefore superior to conventional fraud detection techniques, with a flexible, scalable, and robust means of protecting online transactions and addressing sophisticated fraud techniques.

  PublisherDOI

• Enhanced Cloud Threat Mitigation: Leveraging Hybrid Statistical-Machine Learning Models with GANs, Cat Boost, and MDP

  2025-07-04

  article

  This paper presents a hybrid cloud threat mitigation framework that employs Markov Decision Processes, CatBoost, and Generative Adversarial Networks, in order to enhance cybersecurity in dynamic cloud environments. Traditional defenses remain inadequate, as the cyberthreat landscape has become more complicated both from the context of ransomware, distributed denial-of-service attacks, and polymorphic malware. The platform employs real-time decision making to mitigate uncertainty, state-of-the-art threat classification for accuracy, and synthetic attack data generation for training. This is relevant with current cloud infrastructures to enhance detection accuracy, low false alarm rates, and processing is quick. The model ensures persistent protection with limited interruption by adapting to changing threat patterns. Overall, this effectively tackles significant cloud security issues associated with system responsiveness and threat evolution. Future improvements will focus on integrating explainable artificial intelligence techniques to improve model interpretability, adding blockchain technology for additional trust, and assisting in scaling the solution for large cloud platforms. The proposed methodology helps provide a reliable and intelligent approach to detect and mitigate cloud threats in real time.

  PublisherDOI

• AI-enhanced cloud security monitoring: Detecting advanced persistent threats and intrusions using deep autoencoders and hybrid machine learning techniques

  Global Journal of Engineering and Technology Advances · 2025-03-29 · 2 citations

  articleOpen access

  Cloud computing is slowly becoming one of the main infrastructures for businesses, putting it at risk to undergo Advanced Persistent Threats (APTs) and advanced cyberattacks. Traditional intrusion detection systems (IDS) use rule-based or signature-based techniques, which cannot identify zero-day attacks and evolving threats since they solely depend on predefined attacks' signatures. This study proposes an AI-enhanced continuous security monitoring system that combines deep autoencoders for anomaly detection with a hybrid model, MLP-GRU, for threat classification. The deep autoencoder accurately learns network activity and detects deviations, while the MLP-GRU model analyses sequential data patterns, which leads to the increase in classification accuracy. Experimental results using key performance metrics of accuracy, precision, recall, F1-score, and AUC-ROC confirm the efficiency of the proposed system, ensuring its success in differentiating normal from harmful activity. Besides, the throughput analysis demonstrates that it functions in real time to take care of security events within the system. The proposed methodology serves as a viable alternative to conventional IDSs, enhancing the scalability, adaptability, and accuracy of malware detection. Conclusively, future research will focus on adaptive learning, federated security monitoring, and explainable AI towards realizing enhanced detection capabilities.

  PublisherOA PDFDOI

• Multi-Access Edge Computing and AI: Combining Proxy-Based Coordination, k-NN and Artificial Bee Colony (ABC) for Data Privacy and Threat Classification

  2025-01-01

  articleOpen access1st authorCorresponding
  PublisherDOI

• AI-Powered Dynamic Task Scheduling and Optimization for Robotic Clouds with Chebyshev Chaotic Descent and DRL

  2025-07-23

  article

  The paper presents a smart task allocation and load balancing robotic cloud automation system under uncertain and dynamic conditions, inefficient through traditional techniques. Adaptive and intelligent decision-making is inevitable in this case. Two notable strategies for efficient real-time robotic cloud automation are proposed in this paper: Enhanced Weighted Fuzzy Optimization (E-WFO) for prioritized task allocation and Constrained Causal Deep Reinforcement-Based Model (C2DRBM) for dynamic task distribution to resources. E-WFO prioritizes the allocation of tasks, and C2DRBM distributes resources and optimizes the load balancing within a robotic cloud. The experiment result shows the better outcome with 93% accuracy, 92% precision, and 5.9% improvement in Root Mean Error (RME), validating the effectiveness of the proposed system. The hybrid strategy is ideal for smart task allocation, load balancing, and resource optimization, improving task efficiency and scalability in robotic cloud environments.

  PublisherDOI

• Enhancing Rural E-Commerce Logistics Service Quality Using the Kano-Ahp Framework and Machine Learning-Driven Insights

  2025-08-28

  article

  The persistent issues of rural e-commerce logistics include decentralized operations, inefficient cost structures, and a complete lack of infrastructures in the rural areas. It has become increasingly evident that service gaps can be addressed by the new optimization and machine learning techniques due to traditional service quality models such as SERVQUAL-LSQ being unadaptable for rural-based environments. This study incorporates economy and convenience features to SERVQUAL-LSQ in order to offer a logistics evaluation paradigm oriented toward rural areas. Random Forest Regression is applied for predictive analysis, K-means clustering is performed for segmentation, and Particle Swarm Optimization is used to optimize the weight as an avenue to improve the service quality. The case study measured the perceived and expected service quality with statistical analysis and structured surveys at Suning Tesco. It results in a 20 % increase in customer satisfaction, a 15 % improvement in delivery efficiency, and a <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{2 5 \%}$</tex> reduction in the service quality gap.This framework works far better than traditional methods in attempting to solve logistics issues for rural areas. It bridges some operational gaps and improves customer experience in rural e-commerce logistics, while using optimization and machine learning to ensure scalability, resilience, and quality service.

  PublisherDOI

Frequent coauthors

• Param Vir Singh

  Film Independent

  15 shared

• Shunyuan Zhang

  11 shared

• Zijun Shi

  11 shared

• Sunder Kekre

  8 shared

• Dokyun Lee

  8 shared

• Jian Ni

  Virginia Tech

  7 shared

• Nitin Mehta

  7 shared

• Xiao Liu

  6 shared

Education

• Ph.D., Management

  UCLA Anderson School of Management

  1986