BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Metabolically driven latent space learning for gene expression dat
 a.  A journey through manifolds and Pareto fronts - Marco Barsacchi\, Univ
 ersity of Firenze\, Italy
DTSTART:20180608T150000Z
DTEND:20180608T160000Z
UID:TALK106813@talks.cam.ac.uk
CONTACT:Pietro Lio
DESCRIPTION:Gene expression microarrays provide a characterisation of the 
 transcriptional\nactivity of a particular\nbiological sample. While exteme
 ly informative\, their high dimensionality\nhampers the process of pattern
 \nextraction. Several approaches have been proposed for gleaning informati
 on\nabout the hidden structure of the\ndata. Among these approaches\, deep
  generative models provide a powerful way for\napproximating the manifold\
 non which the data reside.\nIn this talk I will introduce a deep learning 
 based framework that provides\nnovel insight into the\nmanifold learning f
 or gene expression data\, employing a metabolic model to\nconstrain the le
 arned\nrepresentation. The proposed framework is evaluated\, showing its a
 bility to\ncapture biologically relevant\nfeatures\, and encoding that fea
 tures in a much simpler latent space. We showed\nhow using a metabolic mod
 el\nto drive the autoencoder learning process helps in achieving better\ng
 eneralisation to unseen data.\n
LOCATION:Department of Computer Science and technology\, sw01 
END:VEVENT
END:VCALENDAR