Filters








646 Hits in 4.3 sec

Asymptotic Optimality of Transductive Confidence Machine [chapter]

Vladimir Vovk
2002 Lecture Notes in Computer Science  
praktiqeskie vyvody teorii vero tnoste mogut byt obosnovany v kaqestve sledstvi gipotez o predel no pri dannyh ograniqeni h slo nosti izuqaemyh vleni Abstract Transductive Confidence Machine (TCM) is a  ...  way of converting standard machine-learning algorithms into algorithms that output predictive regions rather than point predictions.  ...  111/BIO14428 "Pattern Recognition Techniques for Gene and Promoter Identification and Classification in Plant Genomic Sequences"), and EU (grant IST-1999-10226 "Eu-rEdit: The Development and Evaluation of  ... 
doi:10.1007/3-540-36169-3_27 fatcat:3vwmxim4nbauflw3avzzngya5a

Discussion on Hedging Predictions in Machine Learning by A. Gammerman and V. Vovk

2007 Computer journal  
The second result is that asymptotically the relative number of cases when the real output value is within confidence interval converges to the average value of conformal predictors.  ...  In this case, there are both frequentist [ , so in Section 4 when the authors find an improvement over the 'Bayes-optimal predictor' and talk of a conformal predictor being 'asymptotically as good as  ... 
doi:10.1093/comjnl/bxl066 fatcat:zjwrwxskarcetan6pwoichevv4

Hedging Predictions in Machine Learning

Alexander Gammerman, Vladimir Vovk
2007 Computer journal  
In particular, it becomes possible to control (up to statistical fluctuations) the number of erroneous predictions by selecting a suitable confidence level.  ...  This can be done successfully using the powerful machinery of modern machine learning.  ...  ACKNOWLEDGEMENTS This work is partially supported by MRC (grant 'Proteomic analysis of the human serum proteome') and the Royal Society (grant 'Efficient pseudo-random number generators').  ... 
doi:10.1093/comjnl/bxl065 fatcat:b2c2wytfzbc4lkcqwiazkaigge

Bayesian transduction and Markov conditional mixtures for spatiotemporal interactive segmentation

Noah Lee, Andrew F. Laine, Shahram Ebadollahi, Robert L. DeLaPaz
2009 2009 4th International IEEE/EMBS Conference on Neural Engineering  
In this paper we propose a novel transductive learning machine for spatiotemporal classification casted as an interactive segmentation problem.  ...  We present Markov conditional mixtures of naïve Bayes models with spatiotemporal regularization constraints in a transductive learning and inference framework.  ...  The transductive generative formalism w.r.t. time allows us to provide i) predictive confidence of the classification for non i.i.d. sequential data and ii) assess performance guarantees of the inference  ... 
doi:10.1109/ner.2009.5109274 fatcat:kaydt3adkbd2jlxzpm7jo5a6ti

Bayesian Transductive Markov Random Fields for Interactive Segmentation in Retinal Disorders [chapter]

Noah Lee, Andrew F. Laine, R. Theodore Smith
2009 International Federation for Medical and Biological Engineering Proceedings  
In the realm of computer aided diagnosis (CAD) interactive segmentation schemes have been well received by physicians, where the combination of human and machine intelligence can provide improved segmentation  ...  In this paper we present extended work on Bayesian transduction and regularized conditional mixtures for interactive segmentation [3].  ...  A special case of our model is the transductive Gaussian mixture model or the optimal transductive Bayes classifier.  ... 
doi:10.1007/978-3-642-03891-4_61 fatcat:g4h7zb3yxvh4rb52ck7beq3gdu

Gamification of Pure Exploration for Linear Bandits [article]

Rémy Degenne, Pierre Ménard, Xuedong Shang, Michal Valko
2020 arXiv   pre-print
First, we provide a thorough comparison and new insight over different notions of optimality in the linear case, including G-optimality, transductive optimality from optimal experimental design and asymptotic  ...  Second, we design the first asymptotically optimal algorithm for fixed-confidence pure exploration in linear bandits.  ...  Research Agency as part of the "Investissements d'avenir" program, reference ANR19-P3IA-0001 (PRAIRIE 3IA Institute) and the project BOLD, reference ANR19-CE23-0026-04.  ... 
arXiv:2007.00953v1 fatcat:fr5nptwpobgfdcyjhsotsi7yhu

Open set face recognition using transduction

Fayin Li, H. Wechsler
2005 IEEE Transactions on Pattern Analysis and Machine Intelligence  
Sects. 3 and 4 discuss the learning theory behind transduction. In particular, Sect. 4 describes the TCM (Transduction Confidence Machine) (Proedrou, 2001).  ...  The paper introduces the Open Set TCM -kNN (Transduction Confidence Machine -k Nearest Neighbors), which is a novel realization of transductive inference that is suitable for open set multi-class classification  ...  Towards that end we introduced the Open Set TCM -kNN (Transduction Confidence Machine -k Nearest Neighbors), a novel realization of transductive inference that is suitable for open set multi-class classification  ... 
doi:10.1109/tpami.2005.224 pmid:16285369 fatcat:rtuxfrzm2zew3o63g4hracynha

Self-calibrating Probability Forecasting

Vladimir Vovk, Glenn Shafer, Ilia Nouretdinov
2003 Neural Information Processing Systems  
Remark The notion of VPM is a version of Transductive Confidence Machine (TCM) introduced in [8] and [9], and Theorem 1 is a version of Theorem 1 in [2]. 4 Discussion of the Venn probability machine In  ...  This paper develops the approach of [2–4], which show that it is possible to produce valid, asymptotically optimal, and practically useful p-values; the p-values can be then used for region prediction.  ... 
dblp:conf/nips/VovkSN03 fatcat:ns3m67nymrbdbkye2k5eodi6ua

Choosing Answers in Epsilon-Best-Answer Identification for Linear Bandits

Marc Jourdan, Rémy Degenne
2022 International Conference on Machine Learning  
We demonstrate that picking the answer with highest mean does not allow an algorithm to reach asymptotic optimality in terms of expected sample complexity.  ...  Finally, we propose an asymptotically optimal algorithm for this setting, which is shown to achieve competitive empirical performance against existing modified best-arm identification algorithms.  ...  The question of finding an ε-optimal point of a reward function in a non-finite set is the general question of optimization, which is central to many areas of machine learning.  ... 
dblp:conf/icml/JourdanD22 fatcat:teokwfhaljc2db5loubmidvjay

Choosing Answers in ε-Best-Answer Identification for Linear Bandits [article]

Marc Jourdan, Rémy Degenne
2022 arXiv   pre-print
We demonstrate that picking the answer with highest mean does not allow an algorithm to reach asymptotic optimality in terms of expected sample complexity.  ...  Finally, we propose an asymptotically optimal algorithm for this setting, which is shown to achieve competitive empirical performance against existing modified best-arm identification algorithms.  ...  The question of finding an ε-optimal point of a reward function in a non-finite set is the general question of optimization, which is central to many areas of machine learning.  ... 
arXiv:2206.04456v1 fatcat:u52oqknbijgtzkor4twh5cus34

An overview of statistical learning theory

V.N. Vapnik
1999 IEEE Transactions on Neural Networks  
In the middle of the 1990's new types of learning algorithms (called support vector machines) based on the developed theory were proposed.  ...  Until the 1990's it was a purely theoretical analysis of the problem of function estimation from a given collection of data.  ...  It describes the conditions under which the learning machine implementing ERM principle has an asymptotic high rate of convergence independently of the problem to be solved.  ... 
doi:10.1109/72.788640 pmid:18252602 fatcat:dmvybowcbjfmpc7p4uv7py225a

Cyberspace Security Using Adversarial Learning and Conformal Prediction

Harry Wechsler
2015 Intelligent Information Management  
The motivation for using conformal prediction and its immediate off-spring, those of semi-supervised learning and transduction, comes from them first and foremost supporting discriminative and non-parametric  ...  methods characteristic of principled demarcation using cohorts and sensitivity analysis to hedge on the prediction outcomes including negative selection, on one side, and providing credibility and confidence  ...  Transduction, in general, and both the Transduction Confidence Machine (TCM) and the Transduction Confidence Machine for Detection and Recognition (TCM-DR), in particular, which are examples of CP offspring  ... 
doi:10.4236/iim.2015.74016 fatcat:wqiu3pkl6zeurlr3mizdahhgd4

A novel transductive SVM for semisupervised classification of remote sensing images

Mingmin Chi, Lorenzo Bruzzone, Lorenzo Bruzzone
2005 Image and Signal Processing for Remote Sensing XI  
Index Terms-Ill-posed problems, labeled and unlabeled patterns, machine learning, remote sensing, semisupervised classification, support vector machines (SVMs), transductive inference.  ...  The method is based on recent developments in statistical learning theory concerning transductive inference and in particular transductive SVMs (TSVMs).  ...  value) is the growth rate, which, together with the maximum regularization value, controls the asymptotic convergence of the algorithm.  ... 
doi:10.1117/12.628862 fatcat:e7iv2jnlkrdctjtbft63kdhrli

A Novel Transductive SVM for Semisupervised Classification of Remote-Sensing Images

L. Bruzzone, M. Chi, M. Marconcini
2006 IEEE Transactions on Geoscience and Remote Sensing  
Index Terms-Ill-posed problems, labeled and unlabeled patterns, machine learning, remote sensing, semisupervised classification, support vector machines (SVMs), transductive inference.  ...  The method is based on recent developments in statistical learning theory concerning transductive inference and in particular transductive SVMs (TSVMs).  ...  value) is the growth rate, which, together with the maximum regularization value, controls the asymptotic convergence of the algorithm.  ... 
doi:10.1109/tgrs.2006.877950 fatcat:tl2tslxlcre2jpboi35dhy66ka

Single Point Transductive Prediction [article]

Nilesh Tripuraneni, Lester Mackey
2020 arXiv   pre-print
We then provide non-asymptotic upper bounds on the x_ prediction error of two transductive prediction rules.  ...  We address this question in the context of linear prediction, showing how techniques from semi-parametric inference can be used transductively to combat regularization bias.  ...  On asymptotically optimal confidence regions and tests for high-dimensional models. The Annals of Statistics, 42(3):1166-1202, 2014.  ... 
arXiv:1908.02341v4 fatcat:6gi3qzahf5djtcwckxuldj77ma
« Previous Showing results 1 — 15 out of 646 results