Unimodal regularized neuron stick-breaking for ordinal classification

Liu, Xiaofeng; Fan, Fangfang; Kong, Lingsheng; Diao, Zhihui; Xie, Wanqing; Lu, Jun; You, Jane

doi:10.1016/j.neucom.2020.01.025

Cited by 52 publications

(37 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…and t is a one-hot distribution with t j * = 1(or 2 , there is only one feasible optimal transport plan.…”

Section: Optimal Transport With One-hot Targetmentioning

confidence: 99%

“…Conventionally, the risk minimization in deep learning is based on N -way flat softmax prediction and cross-entropy (CE) loss, where N is the number of categories. However, it can ignore the correlation of different classes and can not discriminate different kinds of misclassification [1,2].…”

Section: Introductionmentioning

confidence: 99%

“…We use i, j interlaced for s and t, since they index the same group of categories 2. We note that softmax cannot strictly guarantee the sum of its outputs to be 1 considering the rounding operation.…”

mentioning

confidence: 99%

See 2 more Smart Citations

Embedding Semantic Hierarchy in Discrete Optimal Transport for Risk Minimization

et al. 2021

ICASSP 2021 - 2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

Self Cite

View full text Add to dashboard Cite

The widely-used cross-entropy (CE) loss-based deep networks achieved significant progress w.r.t. the classification accuracy. However, the CE loss can essentially ignore the risk of misclassification which is usually measured by the distance between the prediction and label in a semantic hierarchical tree. In this paper, we propose to incorporate the risk-aware inter-class correlation in a discrete optimal transport (DOT) training framework by configuring its ground distance matrix. The ground distance matrix can be pre-defined following a priori of hierarchical semantic risk. Specifically, we define the tree induced error (TIE) on a hierarchical semantic tree and extend it to its increasing function from the optimization perspective. The semantic similarity in each level of a tree is integrated with the information gain. We achieve promising results on several large scale image classification tasks with a semantic tree structure in a plug and play manner.

show abstract

“…and t is a one-hot distribution with t j * = 1(or 2 , there is only one feasible optimal transport plan.…”

Section: Optimal Transport With One-hot Targetmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Embedding Semantic Hierarchy in Discrete Optimal Transport for Risk Minimization

et al. 2021

ICASSP 2021 - 2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

Self Cite

View full text Add to dashboard Cite

show abstract

“…Compared to the case of a single image [24,37,31,38,29,26], richer and complementary information can be expected in a set, because the samples are captured from multiple views [36]. However, it also poses several challenges including a) variable number of samples within a set, b) larger inner-set variability than its video-based recognition counterpart, and c) order-less data.…”

Section: Introductionmentioning

confidence: 99%

Permutation-Invariant Feature Restructuring for Correlation-Aware Image Set-Based Recognition

Liu

Guo

et al. 2019

2019 IEEE/CVF International Conference on Computer Vision (ICCV)

Self Cite

View full text Add to dashboard Cite

We consider the problem of comparing the similarity of image sets with variable-quantity, quality and un-ordered heterogeneous images. We use feature restructuring to exploit the correlations of both inner&inter-set images. Specifically, the residual self-attention can effectively restructure the features using the other features within a set to emphasize the discriminative images and eliminate the redundancy. Then, a sparse/collaborative learning-based dependency-guided representation scheme reconstructs the probe features conditional to the gallery features in order to adaptively align the two sets. This enables our framework to be compatible with both verification and open-set identification. We show that the parametric self-attention network and non-parametric dictionary learning can be trained endto-end by a unified alternative optimization scheme, and that the full framework is permutation-invariant. In the numerical experiments we conducted, our method achieves top performance on competitive image set/video-based face recognition and person re-identification benchmarks.

show abstract

“…On the other hand, metric regression methods treat the pose as a continuous numerical value [35,28,32], although the label of pose itself is discrete. As manifested in [36,31,27], learning a regression model using discrete labels will cause over-fitting and exhibit similar or inferior performance compared with classification.…”

Section: Introductionmentioning

confidence: 99%

Conservative Wasserstein Training for Pose Estimation

Liu

Zou

Che

et al. 2019

2019 IEEE/CVF International Conference on Computer Vision (ICCV)

Self Cite

View full text Add to dashboard Cite

This paper targets the task with discrete and periodic class labels (e.g., pose/orientation estimation) in the context of deep learning. The commonly used cross-entropy or regression loss is not well matched to this problem as they ignore the periodic nature of the labels and the class similarity, or assume labels are continuous value. We propose to incorporate inter-class correlations in a Wasserstein training framework by pre-defining (i.e., using arc length of a circle) or adaptively learning the ground metric. We extend the ground metric as a linear, convex or concave increasing function w.r.t. arc length from an optimization perspective. We also propose to construct the conservative target labels which model the inlier and outlier noises using a wrapped unimodal-uniform mixture distribution. Unlike the one-hot setting, the conservative label makes the computation of Wasserstein distance more challenging. We systematically conclude the practical closed-form solution of Wasserstein distance for pose data with either one-hot or conservative target label. We evaluate our method on head, body, vehicle and 3D object pose benchmarks with exhaustive ablation studies. The Wasserstein loss obtaining superior performance over the current methods, especially using convex mapping function for ground metric, conservative label, and closed-form solution.

show abstract

Unimodal regularized neuron stick-breaking for ordinal classification

Cited by 52 publications

References 22 publications

Embedding Semantic Hierarchy in Discrete Optimal Transport for Risk Minimization

Embedding Semantic Hierarchy in Discrete Optimal Transport for Risk Minimization

Permutation-Invariant Feature Restructuring for Correlation-Aware Image Set-Based Recognition

Conservative Wasserstein Training for Pose Estimation

Contact Info

Product

Resources

About