BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:Interpretable Machine Learning
DTSTART:20170510T100000
DTEND:20170510T113000
DTSTAMP:20260411T124420Z
UID:6ec6a39744504cd5f24702e697d7fb8f675836703dedea7fee0e5ca7
CATEGORIES:Conferences - Seminars
DESCRIPTION:Professor Fernando Perez-Cruz\nAssociate Professor in Computer
  Science\nStevens Institute of Technology\nExtended Abstract\n \nMachine 
 learning relies on two general approaches to learning from data: discrimin
 ative and generative modeling. Discriminative techniques focus on solving 
 a clear well-defined supervised learning task. These models use labeled da
 tasets to predict an output given any potential input. On the contrary\, g
 enerative approaches build a probability density model for the data\, and 
 do not have a clear task or metric at hand\, so potentially they could sol
 ve any.\n \nAnother way of thinking about these two approaches is to look
  at discriminative learning as a one-way conversation\, and at generative 
 modeling as a two-way one. The goal of discriminative models is to build a
  machine as accurate as possible\, given some labeled data and a specific 
 metric. Generative modeling\, on the other hand\, requires the interaction
  between data owner (expert in the field) and a machine-learning practitio
 ner (building the model). Generative models should include all prior intan
 gible information known to the data owner and should also be understandabl
 e to the data owner because the purpose of these models is not reducing so
 me error measure but learning and understanding the data. In this sense\,
  generative models need to be either human interpretable and actionable\, 
 and/or point to causal interactions. \n \nIn today’s talk\, we will fi
 rst discuss how Bayesian nonparametric (BNP) models can be used to find hi
 dden patterns in data. BNPs return latent variable models that following D
 e Finetti’s Theorem\, find latent variables for which the observed data 
 are conditional independent. From the analysis of this latent variable mod
 el\, we can gain knowledge about the system that generated the data and dr
 aw actionable conclusions and propose tests to verify the found connection
 s. \n \nThen\, towards the end of the presentation we will focus on the 
 application of BNPs to modeling psychiatric disorders\, geo-localization w
 ith LTE networks and trading relationships. These latent variable models y
 ield not only intuitive\, but perhaps unsurprising findings\, but also rel
 evant interactions not evident to the naked eye.\n 
LOCATION:BC 420 https://plan.epfl.ch/?room==BC%20420
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR