UPCOMING EVENTS

Benefits of Weighted Training in Machine Learning and PDE-based Inverse Problems

2022-11-04
Date: 2022-11-04 16:00:00
Time: 4th Nov. 2022 16:00-17:30pm(Beijing)
Venue: online
Speaker: Yang Yunan
Category: Talk & Lecture

Speaker:Yang Yunan (ETH Zurich)


Avenue: Online

    Tencent Meeting ID: 153-737-210 153 737 210


Abstract: 

Many  models  in  machine  learning  and  PDE-based  inverse  problems  exhibit intrinsic spectral properties, which have been used to explain the generalization  capacity  and  the  ill-posedness  of  such  problems.  In  this  talk,   we   discuss   weighted   training   for   computational   learning   and   inversion  with  noisy  data.  The  highlight  of  the  proposed  framework  is  that  we  allow  weighting  in  both  the  parameter  space  and  the  data  space.  The  weighting  scheme  encodes  both  a  priori  knowledge  of  the  object  to  be learned and a strategy to weight the contribution of training data in the loss  function.  We  demonstrate  that  appropriate  weighting  from  prior  knowledge  can  improve  the  generalization  capability  of  the  learned  model in both machine learning and PDE-based inverse problems.Many  models  in  machine  learning  and  PDE-based  inverse  problems  exhibit intrinsic spectral properties, which have been used to explain the generalization  capacity  and  the  ill-posedness  of  such  problems.  In  this  talk,   we   discuss   weighted   training   for   computational   learning   and   inversion  with  noisy  data.  The  highlight  of  the  proposed  framework  is  that  we  allow  weighting  in  both  the  parameter  space  and  the  data  space.  The  weighting  scheme  encodes  both  a  priori  knowledge  of  the  object  to  be learned and a strategy to weight the contribution of training data in the loss  function.  We  demonstrate  that  appropriate  weighting  from  prior  knowledge  can  improve  the  generalization  capability  of  the  learned  model in both machine learning and PDE-based inverse problems.Many models in machine learning and PDE-based inverse problems exhibit intrinsic spectral properties, which have been used to explain the generalization capacity and the ill-posedness of such problems. In this talk, we discuss weighted training for computational learning and inversion with noisy data. The highlight of the proposed framework is that we allow weighting in both the parameter space and the data space. The weighting scheme encodes both a priori knowledge of the object to be learned and a strategy to weight the contribution of training data in the loss function. We demonstrate that appropriate weighting from prior knowledge can improve the generalization capability of the learned model in both machine learning and PDE-based inverse problems.