Filters








1,701 Hits in 5.8 sec

Calibrated Surrogate Maximization of Linear-fractional Utility in Binary Classification [article]

Han Bao, Masashi Sugiyama
2020 arXiv   pre-print
In this paper, we consider linear-fractional metrics, which are a family of classification performance metrics that encompasses many standard ones such as the F_β-measure and Jaccard index, and propose  ...  We characterize sufficient conditions which make the surrogate maximization coincide with the maximization of the true utility.  ...  Supplementary Material for "Calibrated Surrogate Maximization of Linear-fractional Utility in Binary Classification" A Calibration Analysis and Deferred Proofs from Section 4 In this section, we analyze  ... 
arXiv:1905.12511v2 fatcat:7vujdochqjbzbk2eon3ift2e4m

Convex Calibrated Surrogates for the Multi-Label F-Measure [article]

Mingyuan Zhang, Harish G. Ramaswamy, Shivani Agarwal
2020 arXiv   pre-print
In this paper, we explore the question of designing convex surrogate losses that are calibrated for the F-measure – specifically, that have the property that minimizing the surrogate loss yields (in the  ...  In particular, the F-measure explicitly balances recall (fraction of active labels predicted to be active) and precision (fraction of labels predicted to be active that are actually so), both of which  ...  SA is also supported in part by the US National Institutes of Health (NIH) under Grant No. U01CA214411.  ... 
arXiv:2009.07801v1 fatcat:jfk5shpu75aardsxrg6lxctaiq

Theory of Optimizing Pseudolinear Performance Measures: Application to F-measure [article]

Shameem A Puthiya Parambath, Nicolas Usunier, Yves Grandvalet
2018 arXiv   pre-print
We establish that many notions of F-measures and Jaccard Index are pseudo-linear functions of the per-class false negatives and false positives for binary, multiclass and multilabel classification.  ...  We also establish the multi-objective nature of the F-score maximization problem by linking the algorithm with the weighted-sum approach used in multi-objective optimization.  ...  Acknowledgments This work was carried out and funded in the framework of the Labex MS2T.  ... 
arXiv:1505.00199v4 fatcat:pjwx3fyfjfddpighghpe53frqi

Surrogate regret bounds for generalized classification performance metrics [article]

Wojciech Kotłowski, Krzysztof Dembczyński
2016 arXiv   pre-print
We consider optimization of generalized performance metrics for binary classification by means of surrogate losses.  ...  We focus on a class of metrics, which are linear-fractional functions of the false positive and false negative rates (examples of which include F_β-measure, Jaccard similarity coefficient, AM measure,  ...  Parambath et al (2014) presented an alternative approach to maximizing linear-fractional metrics by learning a sequence of binary classification problems with varying misclassification costs.  ... 
arXiv:1504.07272v2 fatcat:yvabengmjfbfjbhfqzhwjahkoq

Surrogate regret bounds for generalized classification performance metrics

Wojciech Kotłowski, Krzysztof Dembczyński
2016 Machine Learning  
We consider optimization of generalized performance metrics for binary classification by means of surrogate losses.  ...  We focus on a class of metrics, which are linear-fractional functions of the false positive and false negative rates (examples of which include F β -measure, Jaccard similarity coefficient, AM measure,  ...  Parambath et al. (2014) presented an alternative approach to maximizing linear-fractional metrics by learning a sequence of binary classification problems with varying misclassification costs.  ... 
doi:10.1007/s10994-016-5591-7 fatcat:fxoliljzdzbabgbthpp37ybupi

A Review on Multi-Label Learning Algorithms

Min-Ling Zhang, Zhi-Hua Zhou
2014 IEEE Transactions on Knowledge and Data Engineering  
During the past decade, significant amount of progresses have been made towards this emerging machine learning paradigm.  ...  This paper aims to provide a timely review on this area with emphasis on state-of-the-art multi-label learning algorithms.  ...  On the other hand, the number of binary classifiers constructed by Calibrated Label Ranking grows from linear scale to quadratic scale in terms of the number class labels (i.e. q).  ... 
doi:10.1109/tkde.2013.39 fatcat:oqvq3cei4vatdld4j4bqeyc7ry

Review of applications and challenges of quantitative systems pharmacology modeling and machine learning for heart failure

Limei Cheng, Yuchi Qiu, Brian J. Schmidt, Guo-Wei Wei
2021 Journal of Pharmacokinetics and Pharmacodynamics  
They provide data-driven approaches in HF diagnosis and modeling, and offer a novel strategy to inform QSP model development and calibration.  ...  However, various challenges exist in the QSP modeling and clinical characterization of HF. Machine/deep learning (ML/DL) methods have had success in a wide variety of fields and disciplines.  ...  Acknowledgements This work was supported in part by NIH grant GM126189, NSF grants DMS-2052983, DMS-1761320, and IIS-1900473, NASA grant 80NSSC21M0023, Michigan State Foundation, Bristol-Myers Squibb 65109  ... 
doi:10.1007/s10928-021-09785-6 pmid:34637069 pmcid:PMC8837528 fatcat:bylc4yazbvbavapkh4lqafdb3y

Consistent Classification with Generalized Metrics [article]

Xiaoyan Wang, Ran Li, Bowei Yan, Oluwasanmi Koyejo
2019 arXiv   pre-print
We show that the plug-in estimator based on this characterization is consistent and is easily implemented as a post-processing rule.  ...  We propose a framework for constructing and analyzing multiclass and multioutput classification metrics, i.e., involving multiple, possibly correlated multiclass labels.  ...  Fractional-Linear Multiclass Metrics Fractional-Linear metrics are a popular family of classification metrics which include the F-measure.  ... 
arXiv:1908.09057v1 fatcat:i473x56wvbbjjnivdejqmwwive

Constrained Classification and Ranking via Quantiles [article]

Alan Mackey, Xiyang Luo, Elad Eban
2018 arXiv   pre-print
In most machine learning applications, classification accuracy is not the primary metric of interest.  ...  The maximization of many of these metrics can be expressed as a constrained optimization problem, where the constraint is a function of the classifier's predictions.  ...  The same result is known for binary cost-sensitive classification [7] .  ... 
arXiv:1803.00067v1 fatcat:e36mrilp7rhpxbslcm3sslz62u

Risk Guarantees for End-to-End Prediction and Optimization Processes [article]

Nam Ho-Nguyen, Fatma Kılınç-Karzan
2020 arXiv   pre-print
In a computational study on portfolio optimization, fractional knapsack and multiclass classification problems, we compare the optimization performance of using of several prediction loss functions (some  ...  In general, verification of these conditions is a non-trivial task.  ...  We would like to thank the review team for their suggestions that lead to significant improvements in terms of the presentation of the material.  ... 
arXiv:2012.15046v1 fatcat:eujkggwjmncc5otmirkb6dwswq

Binary Classification with Karmic, Threshold-Quasi-Concave Metrics [article]

Bowei Yan, Oluwasanmi Koyejo, Kai Zhong, Pradeep Ravikumar
2018 arXiv   pre-print
Complex performance measures, beyond the popular measure of accuracy, are increasingly being used in the context of binary classification.  ...  In this paper, we advance this understanding of binary classification for complex performance measures by identifying two key properties: a so-called Karmic property, and a more technical threshold-quasi-concavity  ...  Similarly, for the Arithmetic Mean (AM) measure, Menon et al. (2013) design a consistent optimization scheme, based on a balanced classification-calibrated surrogate to AM.  ... 
arXiv:1806.00640v1 fatcat:uttigntla5d4pj247ggkwxsjya

Multi-label optimal margin distribution machine

Zhi-Hao Tan, Peng Tan, Yuan Jiang, Zhi-Hua Zhou
2019 Machine Learning  
The pivotal idea is to maximize the minimum margin of label pairs, which is extended from SVM.  ...  Multi-label support vector machine (Rank-SVM) is a classic and effective algorithm for multi-label classification.  ...  Acknowledgements This research was supported by the National Key R&D Program of China (2018YFB1004300), NSFC (61673201), and the Collaborative Innovation Center of Novel Software Technology and Industrialization  ... 
doi:10.1007/s10994-019-05837-8 fatcat:okilchssg5efvceljvlqljz3ly

Efficient Policy Learning from Surrogate-Loss Classification Reductions [article]

Andrew Bennett, Nathan Kallus
2020 arXiv   pre-print
We consider the estimation problem given by a weighted surrogate-loss classification reduction of policy learning with any score function, either direct, inverse-propensity weighted, or doubly robust.  ...  In light of this, we instead propose an estimation approach based on generalized method of moments, which is efficient for the policy parameters.  ...  For classification, Bartlett et al. (2006) studies which losses are appropriate surrogates, i.e., are classification-calibrated.  ... 
arXiv:2002.05153v1 fatcat:wzvxrvfnpnee7knb6ogidgmczq

Automated pressure map segmentation for quantifying phalangeal kinetics during cylindrical gripping

Erik W. Sinsel, Daniel S. Gloekler, Bryan M. Wimer, Christopher M. Warren, John Z. Wu, Frank L. Buczek
2016 Medical Engineering and Physics  
We present a method fusing six degree-offreedom hand kinematics and a kinematic calibration of a cylinder-wrapped pressure film.  ...  Cylindrical handles wrapped with pressure film grids have been used in studies of gripping kinetics.  ...  The goal in the surrogate trial was to perform a binary classification of pressure cells as being either inside or outside of R A .  ... 
doi:10.1016/j.medengphy.2015.11.004 pmid:26709291 pmcid:PMC4830423 fatcat:5zakbhyl5naytoueshslvka64e

Support Vector Machines for Classification: A Statistical Portrait [chapter]

Yoonkyung Lee
2009 Msphere  
The support vector machine is a supervised learning technique for classification increasingly used in many applications of data mining, engineering, and bioinformatics.  ...  In addition, statistical properties that illuminate both advantage and limitation of the method due to its specific mechanism for classification are briefly discussed.  ...  The solid line in Figure 1 indicates the boundary of the linear SVM classifier with maximal margin for the toy example, and the dotted lines are 1 and −1 level sets of the discriminant function.  ... 
doi:10.1007/978-1-60761-580-4_11 pmid:20652511 fatcat:5dhalqbrxrdtfoggvek36x74im
« Previous Showing results 1 — 15 out of 1,701 results