Knowledge extraction from gene expression data has been one of the main challenges in the bioinformatics field during the last few years. In this context, a particular kind of data, data retrieved in a temporal basis (also known as time series), provide information about the way a gene can be expressed during time. This work presents an exhaustive analysis of last proposals in this area, particularly focusing on those proposals using non-supervised machine learning techniques (i.e. clustering, biclustering and regulatory networks) to find relevant patterns in gene expression. © 2011 Springer-Verlag.