Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Extreme Learning Machines for Fast Training of Click-Through Rate Prediction Models

Ergun Biçici

TL;DR

The findings show that ELMs can be useful for CTR prediction, especially when fast training is needed, and an ELM-based model enhanced with embedding layers is introduced to improve the performance on CTR tasks, which is a novel addition to the field.

Abstract

Extreme Learning Machines (ELM) provide a fast alternative to traditional gradient-based learning in neural networks, offering rapid training and robust generalization capabilities. Its theoretical basis shows its universal approximation capability. We explore the application of ELMs for the task of Click-Through Rate (CTR) prediction, which is largely unexplored by ELMs due to the high dimensionality of the problem. We introduce an ELM-based model enhanced with embedding layers to improve the performance on CTR tasks, which is a novel addition to the field. Experimental results on benchmark datasets, including Avazu and Criteo, demonstrate that our proposed ELM with embeddings achieves competitive F1 results while significantly reducing training time compared to state-of-the-art models such as Masknet. Our findings show that ELMs can be useful for CTR prediction, especially when fast training is needed.

Extreme Learning Machines for Fast Training of Click-Through Rate Prediction Models

TL;DR

The findings show that ELMs can be useful for CTR prediction, especially when fast training is needed, and an ELM-based model enhanced with embedding layers is introduced to improve the performance on CTR tasks, which is a novel addition to the field.

Abstract

Extreme Learning Machines (ELM) provide a fast alternative to traditional gradient-based learning in neural networks, offering rapid training and robust generalization capabilities. Its theoretical basis shows its universal approximation capability. We explore the application of ELMs for the task of Click-Through Rate (CTR) prediction, which is largely unexplored by ELMs due to the high dimensionality of the problem. We introduce an ELM-based model enhanced with embedding layers to improve the performance on CTR tasks, which is a novel addition to the field. Experimental results on benchmark datasets, including Avazu and Criteo, demonstrate that our proposed ELM with embeddings achieves competitive F1 results while significantly reducing training time compared to state-of-the-art models such as Masknet. Our findings show that ELMs can be useful for CTR prediction, especially when fast training is needed.

Paper Structure

This paper contains 9 sections, 2 theorems, 9 equations, 2 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Extreme Learning Machine
  3. Multilayer ELM
  4. Experimental Results
  5. Datasets and Implementation Details
  6. Evaluation Metrics
  7. ELM with Embedding Layer
  8. Performance Evaluation
  9. Conclusion

Key Result

Theorem 1

Given any bounded nonconstant piecewise continuous function $g: \mathbb{R} \rightarrow \mathbb{R}$, if span of $h(\textbf{x}, \tilde{\textbf{w}}, b): (\tilde{\textbf{w}}, b) \in \mathbb{R}^n \times \mathbb{R}$ is dense in $L^2$ space, for any target function $f$ and any function sequence $h_{1 \ldot

Figures (2)

  • Figure 1: ELM: $\tilde{\textbf{W}} \in \mathbf{R}^{N_X \times L}$ and $\tilde{\textbf{b}} \in \mathbf{R}^{N \times 1}$ are random and $\textbf{W} \in \mathbf{R}^{L \times N_Y}$ is learned. Extreme learning machine contains a single hidden layer.
  • Figure 2: ELM-AE contain a single hidden layer.

Theorems & Definitions (2)

  • Theorem 1
  • Theorem 2