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

We show that it is possible to reduce a high-dimensional object like a neuralnetwork agent into a low-dimensional vector representation with semanticmeaning that we call agent embeddings, akin to word or face embeddings. Thiscan be done by collecting examples of existing networks, vectorizing theirweights, and then learning a generative model over the weight space in asupervised fashion. We investigate a pole-balancing task, Cart-Pole, as a casestudy and show that multiple new pole-balancing networks can be generated fromtheir agent embeddings without direct access to training data from theCart-Pole simulator. In general, the learned embedding space is helpful formapping out the space of solutions for a given task. We observe in the case ofCart-Pole the surprising finding that good agents make different decisionsdespite learning similar representations, whereas bad agents make similar (bad)decisions while learning dissimilar representations. Linearly interpolatingbetween the latent embeddings for a good agent and a bad agent yields an agentembedding that generates a network with intermediate performance, where theperformance can be tuned according to the coefficient of interpolation. Linearextrapolation in the latent space also results in performance boosts, up to apoint.

Agent Embeddings: A Latent Representation for Pole-Balancing Networks