BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:TBA
DTSTART:20261120T151500
DTEND:20261120T164500
DTSTAMP:20261107T052611Z
UID:015652fcb1f95a5bc47875f1d2da52e8c59b48d867a2caef4f5fe40c
CATEGORIES:Conferences - Seminars
DESCRIPTION:Johannes Schmidt-Hieber\, WIAS Berlin\nFor small covariate dim
 ension\, shape constrained inference is a well-established topic within no
 nparametric statistics. For large covariate dimensions\, one naturally wan
 ts to introduce machine learning based methods. Input convex neural networ
 ks (ICNNs) were designed as a network architecture to learn convex functio
 ns. In this talk\, we introduce Hyper Input Convex Neural Networks (HyCNNs
 ). HyCNNs combine the principles of Maxout networks with ICNNs to create a
  neural network that is always convex in the input\, theoretically capable
  of leveraging depth\, and performs reliable when trained at scale compare
 d to ICNNs.\nConcretely\, we prove that HyCNNs require exponentially fewer
  parameters than ICNNs to approximate quadratic functions up to a given pr
 ecision. Throughout a series of synthetic experiments\, we demonstrate tha
 t HyCNNs outperform existing ICNNs and MLPs in terms of predictive perform
 ance for convex regression and interpolation tasks. We further apply HyCNN
 s to learn high-dimensional optimal transport maps for synthetic examples 
 and for single-cell RNA sequencing data\, where they oftentimes outperform
  ICNN-based neural optimal transport methods and other baselines across a 
 wide range of settings.\nFor more details\, see arxiv.org/pdf/2604.26942. 
 Joint work with Shayan Hundrieser and Insung Kong
LOCATION:CM 1 517 https://plan.epfl.ch/?room==CM%201%20517
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR