Multi-tiered ML Framework for Digital Campaign Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional digital campaign selection systems employ a random approach, leading to inefficient use of computing and network resources as they transmit campaigns that are unlikely to be interacted with.
Innovation Solution
The system utilizes multiple phases of machine learning models, including propensity models, feature importance models, and student-teacher neural networks, to identify and select digital campaigns that have a high likelihood of being interacted with by users.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If a random approach is used to select digital campaigns for transmission, then the system is simple to implement, but computing and network resources are wasted on transmitting campaigns unlikely to be interacted with
Solution Approach 1:
The system changes the parameter of campaign selection from random to probability-based, using machine learning models to calculate and compare likelihood scores for each campaign. This transforms the selection mechanism to prioritize campaigns with higher expected interaction rates, thereby reducing resource waste while maintaining systematic implementation
Solution Approach 2:
The patent replaces the mechanical random selection process with an intelligent system using propensity models, feature importance models, and student-teacher neural networks. This substitution enables the system to evaluate campaign effectiveness probabilistically, optimizing resource allocation without sacrificing implementation feasibility
2Loss of energy
If multiple machine learning models are used to select digital campaigns, then resource efficiency is improved by transmitting only high-likelihood campaigns, but the system complexity increases
Solution Approach 1:
The system segments the campaign selection process into distinct functional components: propensity models for initial likelihood assessment, feature importance models for identifying key predictive features, and student-teacher neural networks for refined probability estimation. This segmentation allows each component to specialize and operate efficiently, managing overall system complexity through modular architecture
Solution Approach 2:
The feature importance model acts as an intermediary that bridges the propensity model and the student-teacher neural networks. It identifies and prioritizes key features from the input data, reducing the dimensionality and complexity of information processed by subsequent models while maintaining prediction accuracy
Data Source
AI summary
An example method includes receiving a set of features representing attributes for a user and receiving data for multiple digital campaigns. The set of features can be processed using a propensity machine learning model to determine whether the first user is expected to perform an affirmative action in response to the digital campaign. A subset of contributing features that are indicative of a likelihood of the first user performing the affirmative action can be generated using a feature importance model for each digital campaign. For each digital campaign, a respective output indicating a likelihood that the first user will perform an affirmative action in response to the respective digital campaign can be obtained using a respective trained student-teacher neural network. These outputs can be compared to identify a particular digital campaign and one or more digital components associated with this campaign can be transmitted to the first user.