93. Accelerating viral marketing: From combinatorial strategies to learning-based solutions
Invited abstract in session TB-1: Plenary My T. Tai, stream Plenary.
Thursday, 10:00 - 11:00Room: L226
Authors (first author is the speaker)
| 1. | My Thai
|
| CISE Department, University of Florida |
Abstract
The landscape of viral marketing is evolving rapidly, demanding innovative approaches to maximize influence and cost-effectiveness in billion-scale networks. This keynote explores cutting-edge advancements in viral marketing optimization, transcending traditional combinatorial approaches to learning-based strategies. Our journey begins with a traditional (1-1/e)-approximation algorithm with accelerated sampling frameworks, further demonstrating an almost exact solution for this NP-hard problem. Shifting gears to heterogeneous multiplex networks, we introduce combinatorial algorithms that effectively navigate complex overlapping user dynamics, capturing the influences between and within multiple networks. Embracing machine learning, we unveil a transformative framework designed to overcome fundamental obstacles in traditional non-learning-based approaches. Finally, our exploration concludes with MIM-Reasoner, which exploits the reinforcement learning with probabilistic graphical models to address the viral marketing across various platforms.
Keywords
- Machine learning
Status: accepted
Back to the list of papers