Deep Autoencoder Audience Expansion for Real-Time Profile Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional techniques for processing large datasets to expand audiences in targeted marketing are computationally costly and time-consuming, making it difficult to approximate similar user profiles efficiently.
Innovation Solution
Utilizing deep autoencoders to train and generate mathematical representations of seed customers, allowing for real-time expansion of an ideal audience by comparing existing customers with potential viewers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional techniques are used to process large datasets for audience expansion, then audience similarity can be approximated, but computational cost and processing time increase substantially
Solution Approach 1:
The patent segments the large customer dataset into smaller batches that can be processed independently. The autoencoder model processes customers in batches rather than attempting to process the entire dataset at once, which significantly reduces processing time while maintaining the ability to approximate similar user profiles through iterative batch processing
Solution Approach 2:
The patent applies preliminary action by pre-training the autoencoder model on the customer data before actual audience expansion is needed. The model learns the underlying patterns and representations of customer profiles in advance, so that when audience expansion is required, the pre-trained model can quickly process new customers and identify similar profiles without requiring extensive computation at the time of use
2Measurement precision
If conventional techniques are used to process large datasets for audience expansion, then audience similarity can be approximated, but computational cost increases
Solution Approach 1:
The patent extracts only the essential features and patterns from customer data through the autoencoder's latent representation. Instead of processing and comparing entire customer profiles, the system extracts key characteristics into a compressed mathematical representation, which significantly reduces computational cost while preserving the ability to approximate audience similarity
Solution Approach 2:
The patent changes the parameter representation of customer data by transforming raw customer attributes into a different parameter space through the autoencoder. This transformation into latent variables creates a more efficient representation that requires less computational resources for comparison and analysis, thereby reducing overall computational cost
3Productivity
If deep autoencoders are used to generate mathematical representations of customers, then real-time audience expansion is enabled, but system complexity increases
Solution Approach 1:
The patent uses copying by creating a simplified mathematical representation (copy) of customer profiles through the autoencoder's latent space. Instead of working with complex raw customer data, the system creates compressed copies that capture essential characteristics, enabling real-time processing while the underlying model complexity is encapsulated within the pre-trained autoencoder
Data Source
AI summary
A system for selecting an expanded audience comprising: a seed audience; an elastic profile store comprising a plurality of consumer profiles each of the consumer profiles comprising a consumer identifier and consumer characteristics wherein the seed audience is matched in the elastic profile store to select a plurality of seed profiles and at least one candidate profile; at least one encoder engine to encode the seed profiles and the candidate profile to output a plurality of encoded seed profiles and an encoded candidate profile; an aggregator engine to receive the encoded seed profiles to determine similar characteristics; a matching engine to match characteristics from the encoded candidate profile with the characteristics from the encoded seed profiles; a threshold engine to determine whether the encoded candidate profile has sufficient similarity to the encoded seed profiles; and in response to a determination by the threshold engine to keep the candidate profile, inclusion of the candidate user id in an expanded audience.