Artificial Intelligence 13 min read

Optimization of Deep Learning-Based CTR Models in Advertising

This report presents recent advances in optimizing deep learning click‑through‑rate models for advertising, including improved embedding mechanisms, novel feature‑interaction and architecture designs such as attention‑based behavior sequencing, multi‑tower and Mixture‑of‑Experts networks, dynamic ID handling, hourly updates, incremental training, and outlines future multi‑modal and embedding‑importance research.

Ximalaya Technology Team
Ximalaya Technology Team
Ximalaya Technology Team
Optimization of Deep Learning-Based CTR Models in Advertising

This technical report details advancements in optimizing deep learning-based click-through rate (CTR) models for advertising applications. Key improvements include enhanced embedding mechanisms, feature interaction strategies, and model architecture innovations. The study covers sparse and dense component optimizations, dynamic ID feature management, and training framework upgrades. Specific techniques like attention mechanisms for behavioral sequence modeling, multi-tower structures for multi-ad-slot scenarios, and Mixture-of-Experts (MoE) networks for dynamic ad-slot adaptation are discussed. The report also addresses model timeliness improvements through hourly updates and long-term incremental training strategies. Future directions focus on multi-modal content understanding and embedding importance differentiation.

machine learningmodel optimizationdeep learningattention mechanismsAdvertising TechnologyCTR modelEmbedding Techniquesincremental training
Ximalaya Technology Team
Written by

Ximalaya Technology Team

Official account of Ximalaya's technology team, sharing distilled technical experience and insights to grow together.

0 followers
Reader feedback

How this landed with the community

login Sign in to like

Rate this article

Was this worth your time?

Sign in to rate
Discussion

0 Comments

Thoughtful readers leave field notes, pushback, and hard-won operational detail here.