Robust Truncated Hinge Loss Support Vector Machines

By Yichao Wu, Yufeng Liu. 2007

407 citations recorded by the source corpus.

Subfields: Control and Systems Engineering, Statistics and Probability

View Paper: https://www.semanticscholar.org/paper/5a43cc163f4e570b28617ad8a40872ad9189349c

How researchers see this paper

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

In Control and Systems Engineering

In Statistics and Probability

Citations in the corpus

Robust support vector machines based on the rescaled hinge loss function (2017, methodology): “Using the Lagrange multipliers, (13) can be converted into an equivalent dual problem [5,6,13] α α α y α K α C i N 1 min (1)2 − s.”
A Unified View on Multi-class Support Vector Classification (2016, methodology): “This has been observed for particular machines by Guermeur (2007) and by Wu and Liu (2007), but can be shown to be true for all relative margin SVMs. Adding the same vector w to all weight vectors w c (or the same function g ( x ) = (cid:104) w, φ ( x ) (cid:105) to all components f c ) does not…”
Curriculum Loss: Robust Learning and Generalization against Label Corruption (2019, background): “[23] study the truncated hinge loss for SVM.”
Resilient Linear Classification: An Approach to Deal with Attacks on Training Data (2017, positive): “…( e.g., communication channel encryption and authentication), fault tolerant techniques ( e.g., data sanitization [16], robust loss functions [51, 53], and robust learning [14, 20]), and adversarial learning [9, 17, 19] Specifically, the cyber defenses are insufficient for defending against…”
VARIABLE SELECTION IN QUANTILE REGRESSION (2009, methodology): “Liu, Shen and Doss (2005b) and Wu and Liu (2007). DCA is a local algorithm and it decreases the objective value at each iteration. Due to its decomposition and approximation, DCA converges in a finite number of steps. More details on DCA can be found in Liu, Shen and Doss (2005b) and Liu, Shen and W”
Identifying Intention Posts in Discussion Forums (2013, background): “When there are data points far away from their own classes, SVM tends to be strongly affected by such points (Wu & Liu, 2007).”
Are all training examples equally valuable? (2013, background): “For SVMs, the loss function can be changed to become more robust to outliers and bad training examples [20, 5].”
Residual Weighted Learning for Estimating Individualized Treatment Rules (2015, methodology): “The ramp loss (Collobert et al. 2006) is also called the truncated hinge loss function (Wu and Liu 2007).”
The landscape of empirical risk for nonconvex losses (2016, background): “First of all, several empirical studies [4, 21, 15] demonstrate an improved classification error using non-convex losses.”
Enhancing one-class support vector machines for unsupervised anomaly detection (2013, background): “Several approaches have been suggested to tackle this problem [51, 60, 59, 28, 34].”
A Group-Theoretic Framework for Data Augmentation (2019, neutral): “, logistic loss and hinge loss (Wu and Liu, 2007; Nguyen and Sanner, 2013).”
Supervised classification and mathematical optimization (2013, background): “Thus, a classifier more robust against outliers is obtained, [241].”
Algorithms for Direct 0-1 Loss Optimization in Binary Classification (2013, methodology): “…classification methods that are robust to outliers, including tlogistic regression (Ding & Vishwanathan, 2010), robust truncated hinge loss (Wu & Liu, 2007), SavageBoost (Masnadi-Shirazi & Vasconcelos, 2008), and (Collobert et al., 2006) (which shows that a non-convex piecewise linear…”
Stepwise multiple quantile regression estimation using non-crossing constraints (2009, methodology): “This choice was also previously used by Brown et al. (2000) and Wu and Liu (2007).”
Do We Need More Training Data or Better Models for Object Detection? (2012, background): “Interestingly, latent mixture models can mimic the behavior of non-convex bounded loss functions [14] by placing noisy examples into junk”