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

Deep artificial neural networks (DNNs) are typically trained viagradient-based learning algorithms, namely backpropagation. Evolutionstrategies (ES) can rival backprop-based algorithms such as Q-learning andpolicy gradients on challenging deep reinforcement learning (RL) problems.However, ES can be considered a gradient-based algorithm because it performsstochastic gradient descent via an operation similar to a finite-differenceapproximation of the gradient. That raises the question of whethernon-gradient-based evolutionary algorithms can work at DNN scales. Here wedemonstrate they can: we evolve the weights of a DNN with a simple,gradient-free, population-based genetic algorithm (GA) and it performs well onhard deep RL problems, including Atari and humanoid locomotion. The Deep GAsuccessfully evolves networks with over four million free parameters, thelargest neural networks ever evolved with a traditional evolutionary algorithm.These results (1) expand our sense of the scale at which GAs can operate, (2)suggest intriguingly that in some cases following the gradient is not the bestchoice for optimizing performance, and (3) make immediately available themultitude of neuroevolution techniques that improve performance. We demonstratethe latter by showing that combining DNNs with novelty search, which encouragesexploration on tasks with deceptive or sparse reward functions, can solve ahigh-dimensional problem on which reward-maximizing algorithms (e.g.\ DQN, A3C,ES, and the GA) fail. Additionally, the Deep GA is faster than ES, A3C, and DQN(it can train Atari in ${\raise.17ex\hbox{$\scriptstyle\sim$}}$4 hours on onedesktop or ${\raise.17ex\hbox{$\scriptstyle\sim$}}$1 hour distributed on 720cores), and enables a state-of-the-art, up to 10,000-fold compact encodingtechnique.

Deep Neuroevolution: Genetic Algorithms Are a Competitive Alternative for Training Deep Neural Networks for Reinforcement Learning