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 Tools annual

Global: Zendy Free Plan

Global: Zendy Tools monthly

Global: Zendy Plus monthly

In order to apply machine learning classification algorithms, it is assumed that there exists a set of labeled data, called the training set, which is used to train the classifier. However, in real life, this training set may not contain enough labeled data to induce a good classifier. Recently, there has been much interest in a variant of this approach. This new approach, known as semi-supervised learning, assumes that, in addition to the labeled training data, there exists a second set of unlabeled data which is also available during training. One of the goals of semi-supervised learning is training classifiers when large amounts of unlabeled data are available together with a small amount of labeled data. The appeal of semi-supervised learning is due to the fact that in many real-world applications, such sets of unlabeled data are either readily available or inexpensive to collect compared to labeled data. Unlabeled data can often be collected by automated means while labeled data requires human experts or other limited or expensive classification resources that may even be unfeasible in some cases. There are several ways in which unlabeled data can be used. In this work we explore one mechanism by which unlabeled data can be used to improve classification problems and propose a semi-supervised algorithm, called k-meanski, for using unlabeled data for supervised learning. There are two premises underlying the technique used by the proposed algorithm. The first is that input data falls naturally into clusters rather than being uniformly distributed across the entire input space. Furthermore, the initial input labeled data should fall near the center of the existing clusters in the input space. The second premise is that many of the points in some of these clusters belong to specific output categories. Obviously, the validity of these premises is dependent on the dataset used. The k-meanski algorithm works well when the data conform to both premises. If these assumptions are violated poor performance can result. Experiments using real world datasets where we randomly select a subset of the available data to act as labeled exemplars are shown.

"Aprendizado de máquina semi-supervisionado: proposta de um algoritmo para rotular exemplos a partir de poucos exemplos rotulados"