GEN Model: An Alternative Approach to Deep Neural Network Models

TL;DR

The paper presents GEN, a genetic-evolutionary approach to deep learning that constructs a group of shallow models through generations, offering advantages in effectiveness, efficiency, and interpretability over traditional deep models.

Contribution

Introduces GEN, a novel genetic-evolutionary framework for building shallow unit models, differing from traditional deep neural networks, with demonstrated superior performance.

Findings

GEN outperforms state-of-the-art methods in benchmark tests.

GEN is more efficient in training and inference.

GEN offers better interpretability of learned models.

Abstract

In this paper, we introduce an alternative approach, namely GEN (Genetic Evolution Network) Model, to the deep learning models. Instead of building one single deep model, GEN adopts a genetic-evolutionary learning strategy to build a group of unit models generations by generations. Significantly different from the wellknown representation learning models with extremely deep structures, the unit models covered in GEN are of a much shallower architecture. In the training process, from each generation, a subset of unit models will be selected based on their performance to evolve and generate the child models in the next generation. GEN has significant advantages compared with existing deep representation learning models in terms of both learning effectiveness, efficiency and interpretability of the learning process and learned results. Extensive experiments have been done on diverse benchmark datasets, and the experimental results have demonstrated the outstanding performance of GEN compared with the state-of-the-art baseline methods in both effectiveness of efficiency.