English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Hand-crafting effective and efficient structures for recurrent neuralnetworks (RNNs) is a difficult, expensive, and time-consuming process. Toaddress this challenge, we propose a novel neuro-evolution algorithm based onant colony optimization (ACO), called ant swarm neuro-evolution (ASNE), fordirectly optimizing RNN topologies. The procedure selects from multiple modernrecurrent cell types such as Delta-RNN, GRU, LSTM, MGU and UGRNN cells, as wellas recurrent connections which may span multiple layers and/or steps of time.In order to introduce an inductive bias that encourages the formation ofsparser synaptic connectivity patterns, we investigate several variations ofthe core algorithm. We do so primarily by formulating different functions thatdrive the underlying pheromone simulation process (which mimic L1 and L2regularization in standard machine learning) as well as by introducing antagents with specialized roles (inspired by how real ant colonies operate),i.e., explorer ants that construct the initial feed forward structure andsocial ants which select nodes from the feed forward connections tosubsequently craft recurrent memory structures. We also incorporate aLamarckian strategy for weight initialization which reduces the number ofbackpropagation epochs required to locally train candidate RNNs, speeding upthe neuro-evolution process. Our results demonstrate that the sparser RNNsevolved by ASNE significantly outperform traditional one and two layerarchitectures consisting of modern memory cells, as well as the well-known NEATalgorithm. Furthermore, we improve upon prior state-of-the-art results on thetime series dataset utilized in our experiments.

The Ant Swarm Neuro-Evolution Procedure for Optimizing Recurrent Networks