Modeling Latent Variable Uncertainty for Loss-based Learning

By M. Pawan Kumar, Ben Packer, Daphne Koller. 2012

23 citations recorded by the source corpus.

We consider the problem of parameter estimation using weakly supervised datasets, where a training sample consists of the input and a partially specified annotation, which we refer to as the output. The missing information in the annotation is modeled using latent variables. Previous methods overburden a single distribution with two separate tasks: (i) modeling the uncertainty in the latent variables during training; and (ii) making accurate predictions for the output and the latent variables during testing. We propose a novel framework that separates the demands of the two tasks using two distributions: (i) a conditional distribution to model the uncertainty of the latent variables for a given input-output pair; and (ii) a delta distribution to predict the output and the latent variables for a given input. During learning, we encourage agreement between the two distributions by minimizing a loss-based dissimilarity coefficient. Our approach generalizes latent svm in two important ways: (i) it models the uncertainty over latent variables instead of relying on a pointwise estimate; and (ii) it allows the use of loss functions that depend on latent variables, which greatly increases its applicability. We demonstrate the efficacy of our approach on two challenging problems--object detection and action detection--using publicly available datasets.

Subfields: Artificial Intelligence

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

Citations in the corpus

DISCO Nets : DISsimilarity COefficients Networks (2016, methodology): “rsity between P T and P G, and an afﬁne combination of the diversity of each distribution. These coefﬁcients were introduced by Rao [21] with = 1=2 and used for latent variable models in Kumar et al. [10]. We do not need to consider the term DIV (P T;P T) as it is a constant in our problem, and thus”
A survey on still image based human action recognition (2014, methodology): “They demonstrated that the fusion with body parts can improve the action recognition accuracy in their experiments on three datasets: their own Willow Dataset, the Sports Dataset [21] and the PPMI dataset [22].”
On learning to localize objects with minimal supervision (2014, unknown): “Furthermore, smoothing latent SVM implicitly models uncertainty in the selection of the best conﬁgura-tion z in Z , as shown by Kumar et al. (2012) for a different smoothing scheme.”
Dissimilarity Coefficient Based Weakly Supervised Object Detection (2018, methodology): “The aforementioned dissimilarity coefficient based framework has provided promising results in domains where the conditional distribution is simple to model (that is, consists of terms that depend on a few variables at a time) [1, 20].”
Minimum uncertainty latent variable models for robot recognition of sequential human activities (2017, methodology): “In a discriminative setting, CRFs [1], HCRFs [2] and their extensions [3], [5] are the most widely used approaches for modeling activity temporal structures.”