Learning in an Uncertain World: Representing Ambiguity Through Multiple Hypotheses
Authors:
- Christian RupprechtCRChristian Rupprecht
- Iro LainaILIro Laina
- Robert DiPietroRDRobert DiPietro
- Maximilian BaustMBMaximilian Baust
- Federico TombariFTFederico Tombari
- Nassir NavabNNNassir Navab
- Gregory D. HagerGHGregory D. Hager
- 1 October 2017
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 3611-3620
Abstract
Many prediction tasks contain uncertainty. In some cases, uncertainty is inherent in the task itself. In future prediction, for example, many distinct outcomes are equally valid. In other cases, uncertainty arises from the way data is labeled. For example, in object detection, many objects of interest often go unlabeled, and in human pose estimation, occluded joints are often labeled with ambiguous values. In this work we focus on a principled approach for handling such scenarios. In particular, we propose a frame-work for reformulating existing single-prediction models as multiple hypothesis prediction (MHP) models and an associated meta loss and optimization procedure to train them. To demonstrate our approach, we consider four diverse applications: human pose estimation, future prediction, image classification and segmentation. We find that MHP models outperform their single-hypothesis counterparts in all cases, and that MHP models simultaneously expose valuable insights into the variability of predictions.Keywords
All Related Versions
This publication has 18 references indexed in Scilit:
- Fully convolutional networks for semantic segmentation2015
- Anticipating Human Activities Using Object Affordances for Reactive Robotic ResponseIEEE Transactions on Pattern Analysis and Machine Intelligence, 2015
- ImageNet Large Scale Visual Recognition ChallengeInternational Journal of Computer Vision, 2015
- Head Pose Estimation Based on Multivariate Label Distribution2014
- Facial Age Estimation by Learning from Label DistributionsIEEE Transactions on Pattern Analysis and Machine Intelligence, 2013
- A Data-Driven Approach for Event Prediction2010
- The Pascal Visual Object Classes (VOC) ChallengeInternational Journal of Computer Vision, 2009
- Centroidal Voronoi Tessellations: Applications and AlgorithmsSIAM Review, 1999
- Gradient-based learning applied to document recognitionProceedings of the IEEE, 1998
- Least squares quantization in PCMIEEE Transactions on Information Theory, 1982