DeepAI AI Chat
Log In Sign Up

Augmenting Novelty Search with a Surrogate Model to Engineer Meta-Diversity in Ensembles of Classifiers

by   Rui P. Cardoso, et al.
Edinburgh Napier University
Imperial College London

Using Neuroevolution combined with Novelty Search to promote behavioural diversity is capable of constructing high-performing ensembles for classification. However, using gradient descent to train evolved architectures during the search can be computationally prohibitive. Here we propose a method to overcome this limitation by using a surrogate model which estimates the behavioural distance between two neural network architectures required to calculate the sparseness term in Novelty Search. We demonstrate a speedup of 10 times over previous work and significantly improve on previous reported results on three benchmark datasets from Computer Vision – CIFAR-10, CIFAR-100, and SVHN. This results from the expanded architecture search space facilitated by using a surrogate. Our method represents an improved paradigm for implementing horizontal scaling of learning algorithms by making an explicit search for diversity considerably more tractable for the same bounded resources.


page 1

page 2

page 3

page 4


Neural Architecture Search using Progressive Evolution

Vanilla neural architecture search using evolutionary algorithms (EA) in...

NAS-Bench-301 and the Case for Surrogate Benchmarks for Neural Architecture Search

Neural Architecture Search (NAS) is a logical next step in the automatic...

Progressive Neural Architecture Search

We propose a method for learning CNN structures that is more efficient t...

Efficient Architecture Search by Network Transformation

Techniques for automatically designing deep neural network architectures...

One-Shot Neural Ensemble Architecture Search by Diversity-Guided Search Space Shrinking

Despite remarkable progress achieved, most neural architecture search (N...

Traditional and accelerated gradient descent for neural architecture search

In this paper, we introduce two algorithms for neural architecture searc...

Diversity and Novelty MasterPrints: Generating Multiple DeepMasterPrints for Increased User Coverage

This work expands on previous advancements in genetic fingerprint spoofi...