Synthetic Training Data for AI Entity Detection and Ad Frequency Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing advertising systems struggle to optimize and personalize ad delivery across a diverse range of Internet-connected devices, leading to suboptimal user experiences and inefficient ad serving.
Innovation Solution
A system for programmatic generation of training data and entity detection using AI models, combined with advanced frequency management to optimize ad delivery based on entity detection and frequency thresholds.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If AI models are trained on real-world advertising data, then detection accuracy is improved, but data privacy concerns and security risks worsen
Solution Approach 1:
The system creates synthetic advertising data that copies the structural and contextual characteristics of real advertising data without using actual user information. The synthetic data generator produces artificial impressions, clicks, and user interactions that mimic real-world patterns while containing no real personal data, thus eliminating privacy risks while preserving training value.
Solution Approach 2:
The system uses temporary synthetic data that can be generated on-demand and discarded after use, rather than storing and reusing real user data. This disposable approach to data allows repeated training iterations without compromising user privacy, as the synthetic data has no real-world counterpart and can be regenerated as needed.
2Object-affected harmful factors
If frequency management thresholds are strictly enforced, then ad overload is reduced, but ad delivery efficiency worsens
Solution Approach 1:
The frequency management system dynamically adjusts impression thresholds based on real-time contextual factors including time of day, device type, ad relevance score, and user engagement patterns. Rather than using fixed thresholds, the system continuously adapts the frequency limits to balance preventing ad overload with maintaining efficient ad delivery, allowing higher frequencies for high-value ads and lower frequencies for less relevant content.
Solution Approach 2:
The system changes the parameter values of frequency thresholds based on multiple dimensions including temporal patterns, spatial location, content category, and user behavior metrics. By adjusting these parameters dynamically rather than using static limits, the system optimizes the balance between reducing ad fatigue and maintaining delivery efficiency across different contexts.
Data Source
AI summary
Systems and methods for programmatic generation of training data, including: a training data generation engine configured to: identify an image asset corresponding to an entity; identify a training video; select a consecutive subset of frames of the training video based on a procedure for ranking frames on their candidacy for overlaying content; for at least one frame of the subset of frames: perform an augmentation technique on the identified logo image to generate an augmented image asset; overlay at least one variation of the image asset, including the augmented image asset, onto each of the subset of frames to generate a set of overlayed frames; and generate an augmented version of the training video including the overlayed frames; and a model training engine configured to: train an artificial intelligence model for entity detection using the augmented version of the training video.