DOCSLIB.ORG

The Importance of Calibration for Estimating Proportions from Annotations

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
The Importance of Calibration for Estimating Proportions from Annotations The Importance of Calibration for Estimating Proportions from Annotations Dallas Card Noah A. Smith Machine Learning Department Paul G. Allen School of CSE Carnegie Mellon University University of Washington Pittsburgh, PA, 15213, USA Seattle, WA, 98195, USA [email protected] [email protected] Abstract Note that the goal would not be to identify individ- ual instances, but rather to estimate a proportion, Estimating label proportions in a target cor- as a way of measuring the prevalence of a social pus is a type of measurement that is useful for answering certain types of social-scientific phenomenon. Although we assume that trained questions. While past work has described a annotators could recognize this phenomenon with number of relevant approaches, nearly all are some acceptable level of agreement, relying solely based on an assumption which we argue is on manual annotation would restrict the number invalid for many problems, particularly when of messages that could be considered, and would dealing with human annotations. In this paper, limit the analysis to the messages available at the we identify and differentiate between two rele- time of annotation.1 vant data generating scenarios (intrinsic vs. ex- trinsic labels), introduce a simple but novel We thus treat proportion estimation as a mea- method which emphasizes the importance of surement problem, and seek a way to train an in- calibration, and then analyze and experimen- strument from a limited number of human anno- tally validate the appropriateness of various tations to measure label proportions in an unanno- methods for each of the two scenarios. tated target corpus. This problem can be cast within a supervised 1 Introduction learning framework, and past work has demon- A methodological tool often used in the social sci- strated that it is possible to improve upon a na¨ıve ences and humanities (and practical settings like classification-based approach, even without access journalism) is content analysis – the manual cat- to any labeled data from the target corpus (For- egorization of pieces of text into a set of cate- man, 2005, 2008; Bella et al., 2010; Hopkins and gories which have been developed to answer a sub- King, 2010; Esuli and Sebastiani, 2015). How- stantive research question (Krippendorff, 2012). ever, as we argue ( 2), most of this work is based § Automated content analysis holds great promise on a set of assumptions that we believe are in- for augmenting the efforts of human annotators valid in a significant portion of text-based research (O’Connor et al., 2011; Grimmer and Stewart, projects in the social sciences and humanities. 2013). While this task bears similarity to text cat- Our contributions in this paper include: egorization problems such as sentiment analysis, identifying two different data-generating sce- the quantity of real interest is often the proportion • narios for text data (intrinsic vs. extrinsic la- of documents in a dataset that should receive each bels) and and establishing their importance to label (Hopkins and King, 2010). This paper tack- the problem of estimating proportions ( 2); les the problem of estimating label proportions in § a target corpus based on a small sample of human analyzing which methods are suitable for • annotated data. each setting, and proposing a simple alterna- As an example, consider the hypothetical ques- tive approach for extrinsic labels ( 3); and tion (not explored in this work) of whether hate § speech is increasingly prevalent in social media an empirical comparison of methods that val- • posts in recent years. “Hate speech” is a difficult- idates our analysis ( 4). § to-define category only revealed (at least initially) 1For additional examples see Grimmer et al.(2012), Hop- through human judgments (Davidson et al., 2017). kins and King(2010), and references therein. 1636 Proceedings of NAACL-HLT 2018, pages 1636–1646 New Orleans, Louisiana, June 1 - 6, 2018. c 2018 Association for Computational Linguistics Complicating matters somewhat is the fact that controlled real-world process, such as users rat- annotation may take place before the entire col- ing a review’s helpfulness. Note that our setup lection is available, so that the subset of instances does include the special case in which true gold- that are manually annotated may represent a bi- standard labels are available for each instance ased sample ( 2). Because this is so frequently the (such as the authors of documents in an author- § case, all of the results in this paper assume that we ship attribution problem). In such a case, is de- A must confront the challenges of transfer learning terministic (assuming unique inputs). or domain adaptation. (The simpler case, where Given that our objective is to mimic the annota- we can sample from the true population of inter- tion process, we seek to estimate the proportion of est, is revisited in 5.) documents in the target corpus expected to be cat- § egorized into each of the K categories, if we had 2 Problem Definition an unlimited budget and full access to the target Our setup is similar to that faced in transfer learn- corpus at the time of annotation. That is, we wish (T ) ing, and we use similar terminology (Pan and to estimate q , which we define as: Yang, 2010; Weiss et al., 2016). We assume that (T ) 1 NT (T ) q(y = k X ) , p(yi = k x ). we have a source and a target corpus, comprised | NT i=1 | i (2) of NS and NT documents respectively, the latter P of which are not available for annotation. We will Given a set of documents sampled from the represent each corpus as a set of documents, i.e., source corpus and L applications of the annotation (S) (S) X(S) = x , ..., x , and similarly for X(T ). function, we can obtain, at some cost, a labeled h 1 Ns i We further assume that we have a set of K mu- training corpus of L documents, i.e., D(train) = tually exclusive categories, = 1,...,K , and (x , y ),..., (x , y ) . Because the source and Y { } h 1 1 L L i that we wish to estimate the proportion of doc- target corpora are not in general drawn from the uments in the target corpus that belong to each same distribution, we seek to make explicit our as- category. These would typically correspond to a sumptions about how they differ.3 Past literature quantity we wish to measure, such as what frac- on transfer learning has identified several patterns tion of news articles frame a policy issue in a par- of dataset shift (Storkey, 2009). Here we focus ticular way, what fraction of product reviews are on two particularly important cases, linking them considered helpful, or what fraction of social me- to the relevant data generating processes, and ana- dia messages convey positive sentiment. Gener- lyze their relevance to estimating proportions. ally speaking, these categories will be designed based on theoretical assumptions, an understand- Two kinds of distributional shift. There are ing of the design of the platform that produced the two natural assumptions we could make about data, and/or initial exploration of the data itself. what is constant between the two corpora. We In idealized text classification scenarios, it is could assume that there is no change in the dis- conventional to assume training data with already- tribution of text given a document’s label, that is p(S)(x y) = p(T )(x y). Alternately, we could assigned gold-standard labels. Here, we are in- | | assume that there is no change in the distribution terested in scenarios where we must generate our (S) (T ) 2 of labels given text, i.e., p (y x) = p (y labels via an annotation process. Specifically, as- | | sume that we have some annotation function, , x). The former is assumed in the case of prior A which produces a distribution over the K mutu- probability shift, where we assume that p(y) dif- fers but p(x y) is constant, and the later is as- ally exclusive labels, conditional on text. Given | sumed in the case of covariate shift, where we as- a document, xi, the annotation process samples a sume that p(x) differs but p(y x) is constant label from the annotation function, defined as: | (Storkey, 2009). (x , k) p(y = k x ). (1) These two assumptions correspond to two fun- A i , i | i damentally different types of scenarios that we Typically, the annotation function would repre- need to consider, which are summarized in Table sent the behavior of a human annotator (or group 1. The first is where we are dealing with what we of annotators), but it could also represent a less 3Clearly, if we make no assumptions about how the source 2This could include gathering multiple independent anno- and target distributions are related, there is no guarantee that tations per instance, but we will typically assume only one. supervised learning will work (Ben-David et al., 2012). 1637 Label type Intrinsic Extrinsic by a change in the underlying distribution of text, Data generating x p(x y) y p(y x) process ∼ | ∼ | p(x). With intrinsic labels, by contrast, it may be Assumed to differ the case that p(x y) is the same for the source p(y) p(x) | across domains and the target, assuming there are no additional Assumed constant p(x y) p(y x) across domains | | factors influencing the generation of text. In that Corresponding Prior Covariate case, a shift in the distribution of features would distributional probability shift shift shift be fully explained by a difference in the underly- ing label proportions. Table 1: Data generating scenarios and corresponding The idea that there are different data generat- distributional properties.
Load more

Recommended publications
  • Obtaining Well Calibrated Probabilities Using Bayesian Binning
    Proceedings of the Twenty-Ninth AAAI Conference on Artificial Intelligence Obtaining Well Calibrated Probabilities Using Bayesian Binning Mahdi Pakdaman Naeini1, Gregory F. Cooper1;2, and Milos Hauskrecht1;3 1Intelligent Systems Program, University of Pittsburgh, PA, USA 2Department of Biomedical Informatics, University of Pittsburgh, PA, USA 3Computer Science Department, University of Pittsburgh, PA, USA Abstract iments to perform), medicine (e.g., deciding which therapy Learning probabilistic predictive models that are well cali- to give a patient), business (e.g., making investment deci- brated is critical for many prediction and decision-making sions), and others. However, model calibration and the learn- tasks in artificial intelligence. In this paper we present a new ing of well-calibrated probabilistic models have not been non-parametric calibration method called Bayesian Binning studied in the machine learning literature as extensively as into Quantiles (BBQ) which addresses key limitations of ex- for example discriminative machine learning models that isting calibration methods. The method post processes the are built to achieve the best possible discrimination among output of a binary classification algorithm; thus, it can be classes of objects. One way to achieve a high level of model readily combined with many existing classification algo- calibration is to develop methods for learning probabilistic rithms. The method is computationally tractable, and em- models that are well-calibrated, ab initio. However, this ap- pirically accurate, as evidenced by the set of experiments proach would require one to modify the objective function reported here on both real and simulated datasets. used for learning the model and it may increase the com- putational cost of the associated optimization task.
    [Show full text]
  • Calibrated Model-Based Deep Reinforcement Learning
    Calibrated Model-Based Deep Reinforcement Learning and intuition on how to apply calibration in reinforcement Probabilistic Models This paper focuses on probabilistic learning. dynamics models T (s0|s, a) that take a current state s 2S and action a 2A, and output a probability distribution over We validate our approach on benchmarks for contextual future states s0. Web represent the output distribution over the bandits and continuous control (Li et al., 2010; Todorov next states, T (·|s, a), as a cumulative distribution function et al., 2012), as well as on a planning problem in inventory F : S![0, 1], which is defined for both discrete and management (Van Roy et al., 1997). Our results show that s,a continuous Sb. calibration consistently improves the cumulative reward and the sample complexity of model-based agents, and also en- hances their ability to balance exploration and exploitation 2.2. Calibration, Sharpness, and Proper Scoring Rules in contextual bandit settings. Most interestingly, on the A key desirable property of probabilistic forecasts is calibra- HALFCHEETAH task, our system achieves state-of-the-art tion. Intuitively, a transition model T (s0|s, a) is calibrated if performance, using 50% fewer samples than the previous whenever it assigns a probability of 0.8 to an event — such leading approach (Chua et al., 2018). Our results suggest as a state transition (s, a, s0) — thatb transition should occur that calibrated uncertainties have the potential to improve about 80% of the time. model-based reinforcement learning algorithms with mini- 0 mal computational and implementation overhead. Formally, for a discrete state space S and when s, a, s are i.i.d.
    [Show full text]
  • Offline Deep Models Calibration with Bayesian Neural Networks
    Under review as a conference paper at ICLR 2019 OFFLINE DEEP MODELS CALIBRATION WITH BAYESIAN NEURAL NETWORKS Anonymous authors Paper under double-blind review ABSTRACT In this work the authors show that Bayesian Neural Networks (BNNs) can be efficiently applied to calibrate state-of-the-art Deep Neural Networks (DNN). Our approach acts offline, i.e., it is decoupled from the training of the DNN to be cali- brated. This offline approach allow us to apply our BNN calibration to any model regardless of the limitations that the model may present during training. Note that this offline setting is also appropriate in order to deal with privacy concerns of DNN training data or implementation, among others. We show that our approach clearly outperforms other simple maximum likelihood based solutions that have recently shown very good performance, as temperature scaling (Guo et al., 2017). As an example, we reduce the Expected Calibration Error (ECE%) from 0.52 to 0.24 on CIFAR-10 and from 4.28 to 2.46 on CIFAR-100 on two Wide ResNet with 96.13% and 80.39% accuracy respectively, which are among the best results published for these tasks. Moreover, we show that our approach improves the per- formance of online methods directly applied to the DNN, e.g. Gaussian processes or Bayesian Convolutional Neural Networks. Finally, this decoupled approach allows us to apply any further improvement to the BNN without considering the computational restrictions imposed by the deep model. In this sense, this offline setting is a practical application where BNNs can be considered, which is one of the main criticisms to these techniques.
    [Show full text]
  • Discovering General-Purpose Active Learning Strategies
    1 Discovering General-Purpose Active Learning Strategies Ksenia Konyushkova, Raphael Sznitman, and Pascal Fua, Fellow, IEEE Abstract—We propose a general-purpose approach to discovering active learning (AL) strategies from data. These strategies are transferable from one domain to another and can be used in conjunction with many machine learning models. To this end, we formalize the annotation process as a Markov decision process, design universal state and action spaces and introduce a new reward function that precisely model the AL objective of minimizing the annotation cost. We seek to find an optimal (non-myopic) AL strategy using reinforcement learning. We evaluate the learned strategies on multiple unrelated domains and show that they consistently outperform state-of-the-art baselines. Index Terms—Active learning, meta-learning, Markov decision process, reinforcement learning. F 1 INTRODUCTION ODERN supervised machine learning (ML) methods is still greedy and therefore could be a subject to finding M require large annotated datasets for training purposes suboptimal solutions. and the cost of producing them can easily become pro- Our goal is therefore to devise a general-purpose data- hibitive. Active learning (AL) mitigates the problem by se- driven AL method that is non-myopic and applicable to lecting intelligently and adaptively a subset of the data to be heterogeneous datasets. To this end, we build on earlier annotated. To do so, AL typically relies on informativeness work that showed that AL problems could be naturally measures that identify unlabelled data points whose labels reworded in Markov Decision Process (MDP) terms [4], [10]. are most likely to help to improve the performance of the In a typical MDP, an agent acts in its environment.
    [Show full text]
  • On Calibration of Modern Neural Networks
    On Calibration of Modern Neural Networks Chuan Guo * 1 Geoff Pleiss * 1 Yu Sun * 1 Kilian Q. Weinberger 1 Abstract LeNet (1998) ResNet (2016) CIFAR-100 CIFAR-100 1:0 Confidence calibration – the problem of predict- ing probability estimates representative of the 0:8 true correctness likelihood – is important for 0:6 classification models in many applications. We Accuracy Accuracy discover that modern neural networks, unlike 0:4 those from a decade ago, are poorly calibrated. Avg. confidence Avg. confidence Through extensive experiments, we observe that % of Samples 0:2 depth, width, weight decay, and Batch Normal- 0:0 ization are important factors influencing calibra- 0:0 0:2 0:4 0:6 0:8 1:0 0:0 0:2 0:4 0:6 0:8 1:0 tion. We evaluate the performance of various 1:0 Outputs Outputs post-processing calibration methods on state-of- 0:8 Gap Gap the-art architectures with image and document classification datasets. Our analysis and exper- 0:6 iments not only offer insights into neural net- 0:4 work learning, but also provide a simple and Accuracy straightforward recipe for practical settings: on 0:2 Error=44.9 Error=30.6 most datasets, temperature scaling – a single- 0:0 parameter variant of Platt Scaling – is surpris- 0:0 0:2 0:4 0:6 0:8 1:0 0:0 0:2 0:4 0:6 0:8 1:0 ingly effective at calibrating predictions. Confidence Figure 1. Confidence histograms (top) and reliability diagrams (bottom) for a 5-layer LeNet (left) and a 110-layer ResNet (right) 1.
    [Show full text]
  • Calibration Techniques for Binary Classification Problems
    Calibration Techniques for Binary Classification Problems: A Comparative Analysis Alessio Martino a, Enrico De Santis b, Luca Baldini c and Antonello Rizzi d Department of Information Engineering, Electronics and Telecommunications, University of Rome ”La Sapienza”, Via Eudossiana 18, 00184 Rome, Italy Keywords: Calibration, Classification, Supervised Learning, Support Vector Machine, Probability Estimates. Abstract: Calibrating a classification system consists in transforming the output scores, which somehow state the con- fidence of the classifier regarding the predicted output, into proper probability estimates. Having a well- calibrated classifier has a non-negligible impact on many real-world applications, for example decision mak- ing systems synthesis for anomaly detection/fault prediction. In such industrial scenarios, risk assessment is certainly related to costs which must be covered. In this paper we review three state-of-the-art calibration techniques (Platt’s Scaling, Isotonic Regression and SplineCalib) and we propose three lightweight proce- dures based on a plain fitting of the reliability diagram. Computational results show that the three proposed techniques have comparable performances with respect to the three state-of-the-art approaches. 1 INTRODUCTION tion from X . The seminal example is data clustering, where aim of the learning system is to return groups Classification is one of the most important problems (clusters) of data in such a way that patterns belong- falling under the machine learning and, specifically, ing to the same cluster are more similar with respect to under the supervised learning umbrella. Generally patterns belonging to other clusters (Jain et al., 1999; speaking, it is possible to sketch three main families: Martino et al., 2017b; Martino et al., 2018b; Martino clustering, regression/function approximation, classi- et al., 2019; Di Noia et al., 2019).
    [Show full text]
  • Radar-Detection Based Classification of Moving Objects Using Machine Learning Methods
    Radar-detection based classification of moving objects using machine learning methods VICTOR NORDENMARK ADAM FORSGREN Master of Science Thesis Stockholm, Sweden 2015 Radar-detection based classification of moving objects using machine learning methods Victor Nordenmark Adam Forsgren Master of Science Thesis MMK 2015:77 MDA 520 KTH Industrial Engineering and Management Machine Design SE-100 44 STOCKHOLM ii Examensarbete MMK 2015:77 MDA 520 Radar-detection based classification of moving objects using machine learning methods Victor Nordenmark Adam Forsgren Godkänt Examinator Handledare 2015-06-17 Martin Grimheden De-Jiu Chen Uppdragsgivare Kontaktperson Scania Södertälje AB Kristian Lundh Sammanfattning I detta examensarbete undersöks möjligheten att klassificera rörliga objekt baserat på data från Dopplerradardetektioner. Slutmålet är ett system som använder billig hårdvara och utför beräkningar av låg komplexitet. Scania, företaget som har beställt detta projekt, är intresserat av användningspotentialen för ett sådant system i applikationer för autonoma fordon. Specifikt vill Scania använda klassinformationen för att lättare kunna följa rörliga objekt, vilket är en väsentlig färdighet för en autonomt körande lastbil. Objekten delas in i fyra klasser: fotgängare, cyklist, bil och lastbil. Indatan till systemet består väsentligen av en plattform med fyra stycken monopulsdopplerradars som arbetar med en vågfrekvens på 77 GHz. Ett klassificeringssystem baserat på maskininlärningskonceptet Support vector machines har skapats. Detta system har tränats och validerats på ett dataset som insamlats för projektet, innehållandes datapunkter med klassetiketter. Ett antal stödfunktioner till detta system har också skapats och testats. Klassificeraren visas kunna skilja väl på de fyra klasserna i valideringssteget. Simuleringar av det kompletta systemet gjort på inspelade loggar med radardata visar lovande resultat, även i situationer som inte finns representerade i träningsdatan.
    [Show full text]
  • Learning to Plan with Portable Symbols
    Learning to Plan with Portable Symbols Steven James 1 Benjamin Rosman 1 2 George Konidaris 3 Abstract the state space remains complex and planning continues to We present a framework for autonomously learn- be challenging. ing a portable symbolic representation that de- On the other hand, the classical planning approach is to rep- scribes a collection of low-level continuous en- resent the world using abstract symbols, with actions repre- vironments. We show that abstract representa- sented as operators that manipulate these symbols (Ghallab tions can be learned in a task-independent space et al., 2004). Such representations use only the minimal specific to the agent that, when combined with amount of state information necessary for task-level plan- problem-specific information, can be used for ning. This is appealing since it mitigates the issue of reward planning. We demonstrate knowledge transfer sparsity and admits solutions to long-horizon tasks, but in a video game domain where an agent learns raises the question of how to build the appropriate abstract portable, task-independent symbolic rules, and representation of a problem. This is often resolved manu- then learns instantiations of these rules on a per- ally, requiring substantial effort and expertise. Fortunately, task basis, reducing the number of samples re- recent work demonstrates how to learn a provably sound quired to learn a representation of a new task. symbolic representation autonomously, given only the data obtained by executing the high-level actions available to the agent (Konidaris et al., 2018). 1. Introduction A major shortcoming of that framework is the lack of On the surface, the learning and planning communities op- generalisability—an agent must relearn the appropriate sym- erate in very different paradigms.
    [Show full text]
  • Materials Genomics and Machine Learning Kevin Maik Jablonka, Daniele Ongari, Seyed Mohamad Moosavi, and Berend Smit*
    This is an open access article published under an ACS AuthorChoice License, which permits copying and redistribution of the article or any adaptations for non-commercial purposes. pubs.acs.org/CR Review Big-Data Science in Porous Materials: Materials Genomics and Machine Learning Kevin Maik Jablonka, Daniele Ongari, Seyed Mohamad Moosavi, and Berend Smit* Cite This: Chem. Rev. 2020, 120, 8066−8129 Read Online ACCESS Metrics & More Article Recommendations ABSTRACT: By combining metal nodes with organic linkers we can potentially synthesize millions of possible metal−organic frameworks (MOFs). The fact that we have so many materials opens many exciting avenues but also create new challenges. We simply have too many materials to be processed using conventional, brute force, methods. In this review, we show that having so many materials allows us to use big-data methods as a powerful technique to study these materials and to discover complex correlations. The first part of the review gives an introduction to the principles of big-data science. We show how to select appropriate training sets, survey approaches that are used to represent these materials in feature space, and review different learning architectures, as well as evaluation and interpretation strategies. In the second part, we review how the different approaches of machine learning have been applied to porous materials. In particular, we discuss applications in the field of gas storage and separation, the stability of these materials, their electronic properties, and their synthesis. Given the increasing interest of the scientific community in machine learning, we expect this list to rapidly expand in the coming years.
    [Show full text]
  • Measuring Calibration in Deep Learning
    Under review as a conference paper at ICLR 2020 MEASURING CALIBRATION IN DEEP LEARNING Anonymous authors Paper under double-blind review ABSTRACT Overconfidence and underconfidence in machine learning classifiers is measured by calibration: the degree to which the probabilities predicted for each class match the accuracy of the classifier on that prediction. We propose new measures for calibration, the Static Calibration Error (SCE) and Adaptive Calibration Error (ACE). These measures taks into account every prediction made by a model, in contrast to the popular Expected Calibration Error. 1 INTRODUCTION The reliability of a machine learning model’s confidence in its predictions is critical for high risk applications, such as deciding whether to trust a medical diagnosis prediction (Crowson et al., 2016; Jiang et al., 2011; Raghu et al., 2018). One mathematical formulation of the reliability of confidence is calibration (Murphy & Epstein, 1967; Dawid, 1982). Intuitively, for class predictions, calibration means that if a model assigns a class with 90% probability, that class should appear 90% of the time. Calibration is not directly measured by proper scoring rules like negative log likelihood or the quadratic loss. The reason that we need to assess the quality of calibrated probabilities is that presently, we optimize against a proper scoring rule and our models often yield uncalibrated probabilities. Recent work proposed Expected Calibration Error (ECE; Naeini et al., 2015), a measure of calibration error which has lead to a surge of works developing methods for calibrated deep neural networks (e.g., Guo et al., 2017; Kuleshov et al., 2018). In this paper, we show that ECE has numerous pathologies, and that recent calibration methods, which have been shown to successfully recalibrate models according to ECE, cannot be properly evaluated via ECE.
    [Show full text]
  • Arxiv:1808.00111V2 [Cs.LG] 14 Sep 2018
    JMLR: Workshop and Conference Proceedings 77:145{160, 2017 ACML 2017 Probability Calibration Trees Tim Leathart [email protected] Eibe Frank [email protected] Geoffrey Holmes [email protected] Department of Computer Science, University of Waikato Bernhard Pfahringer [email protected] Department of Computer Science, University of Auckland Editors: Yung-Kyun Noh and Min-Ling Zhang Abstract Obtaining accurate and well calibrated probability estimates from classifiers is useful in many applications, for example, when minimising the expected cost of classifications. Ex- isting methods of calibrating probability estimates are applied globally, ignoring the po- tential for improvements by applying a more fine-grained model. We propose probability calibration trees, a modification of logistic model trees that identifies regions of the input space in which different probability calibration models are learned to improve performance. We compare probability calibration trees to two widely used calibration methods|isotonic regression and Platt scaling|and show that our method results in lower root mean squared error on average than both methods, for estimates produced by a variety of base learners. Keywords: Probability calibration, logistic model trees, logistic regression, LogitBoost 1. Introduction In supervised classification, assuming uniform misclassification costs, it is sufficient to pre- dict the most likely class for a given test instance. However, in some applications, it is important to produce an accurate probability distribution over the classes for each exam- ple. While most classifiers can produce a probability distribution for a given test instance, these probabilities are often not well calibrated, i.e., they may not be representative of the true probability of the instance belonging to a particular class.
    [Show full text]
  • Multi-Class Probabilistic Classification Using Inductive and Cross Venn–Abers Predictors
    Proceedings of Machine Learning Research 60:1{13, 2017 Conformal and Probabilistic Prediction and Applications Multi-class probabilistic classification using inductive and cross Venn{Abers predictors Valery Manokhin [email protected] Royal Holloway, University of London, Egham, Surrey, UK Editor: Alex Gammerman, Vladimir Vovk, Zhiyuan Luo and Harris Papadopoulos Abstract Inductive (IVAP) and cross (CVAP) Venn{Abers predictors are computationally efficient algorithms for probabilistic prediction in binary classification problems. We present a new approach to multi-class probability estimation by turning IVAPs and CVAPs into multi- class probabilistic predictors. The proposed multi-class predictors are experimentally more accurate than both uncalibrated predictors and existing calibration methods. Keywords: Probabilistic classification, uncertainty quantification, calibration method, on-line compression modeling, measures of confidence 1. Introduction Multi-class classification is the problem of classifying objects into one of the more than two classes. The goal of classification is to construct a classifier which, given a new test object, will predict the class label from the set of possible k classes. In an \ordinary" classification problem, the goal is to minimize the loss function by predicting correct labels on the test set. In contrast, a probabilistic classifier outputs, for a given test object, the probability distribution over the set of k classes. Probabilistic classifiers allow to express a degree of confidence about the classification of the test object. This is useful in a range of applications and industries, including life sciences, robotics and artificial intelligence. A number of techniques are available to solve binary classification problems. In logistic regression, the dependent variable is a categorical variable indicating one of the two classes.
    [Show full text]