Hidden Markov Models with Confidence. / Cherubin, Giovanni; Nouretdinov, Ilia.

Conformal and Probabilistic Prediction with Applications: 5th International Symposium, COPA 2016 Madrid, Spain, April 20–22, 2016 Proceedings. Vol. 9653 Springer, 2016. p. 128-144 (Lecture Notes in Computer Science; Vol. 9653).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Published

Standard

Hidden Markov Models with Confidence. / Cherubin, Giovanni; Nouretdinov, Ilia.

Conformal and Probabilistic Prediction with Applications: 5th International Symposium, COPA 2016 Madrid, Spain, April 20–22, 2016 Proceedings. Vol. 9653 Springer, 2016. p. 128-144 (Lecture Notes in Computer Science; Vol. 9653).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Harvard

Cherubin, G & Nouretdinov, I 2016, Hidden Markov Models with Confidence. in Conformal and Probabilistic Prediction with Applications: 5th International Symposium, COPA 2016 Madrid, Spain, April 20–22, 2016 Proceedings. vol. 9653, Lecture Notes in Computer Science, vol. 9653, Springer, pp. 128-144. https://doi.org/10.1007/978-3-319-33395-3_10

APA

Cherubin, G., & Nouretdinov, I. (2016). Hidden Markov Models with Confidence. In Conformal and Probabilistic Prediction with Applications: 5th International Symposium, COPA 2016 Madrid, Spain, April 20–22, 2016 Proceedings (Vol. 9653, pp. 128-144). (Lecture Notes in Computer Science; Vol. 9653). Springer. https://doi.org/10.1007/978-3-319-33395-3_10

Vancouver

Cherubin G, Nouretdinov I. Hidden Markov Models with Confidence. In Conformal and Probabilistic Prediction with Applications: 5th International Symposium, COPA 2016 Madrid, Spain, April 20–22, 2016 Proceedings. Vol. 9653. Springer. 2016. p. 128-144. (Lecture Notes in Computer Science). https://doi.org/10.1007/978-3-319-33395-3_10

Author

Cherubin, Giovanni ; Nouretdinov, Ilia. / Hidden Markov Models with Confidence. Conformal and Probabilistic Prediction with Applications: 5th International Symposium, COPA 2016 Madrid, Spain, April 20–22, 2016 Proceedings. Vol. 9653 Springer, 2016. pp. 128-144 (Lecture Notes in Computer Science).

BibTeX

@inproceedings{4709888a3c7b4d9ca2548cad0499e8ee,
title = "Hidden Markov Models with Confidence",
abstract = "We consider the problem of training a Hidden Markov Model(HMM) from fully observable data and predicting the hidden states of anobserved sequence. Our attention is focused to applications that requirea list of potential sequences as a prediction. We propose a novel methodbased on Conformal Prediction (CP) that, for an arbitrary confidencelevel 1 − ε, produces a list of candidate sequences that contains the correctsequence of hidden states with probability at least 1−ε. We presentexperimental results that confirm this holds in practice. We compare ourmethod with the standard approach (i.e.: the use of Maximum Likelihoodand the List–Viterbi algorithm), which suffers from violations tothe assumed distribution. We discuss advantages and limitations of ourmethod, and suggest future directions.",
author = "Giovanni Cherubin and Ilia Nouretdinov",
year = "2016",
month = "4",
day = "17",
doi = "10.1007/978-3-319-33395-3_10",
language = "English",
isbn = "978-3-319-33394-6",
volume = "9653",
series = "Lecture Notes in Computer Science",
publisher = "Springer",
pages = "128--144",
booktitle = "Conformal and Probabilistic Prediction with Applications",

}

RIS

TY - GEN

T1 - Hidden Markov Models with Confidence

AU - Cherubin, Giovanni

AU - Nouretdinov, Ilia

PY - 2016/4/17

Y1 - 2016/4/17

N2 - We consider the problem of training a Hidden Markov Model(HMM) from fully observable data and predicting the hidden states of anobserved sequence. Our attention is focused to applications that requirea list of potential sequences as a prediction. We propose a novel methodbased on Conformal Prediction (CP) that, for an arbitrary confidencelevel 1 − ε, produces a list of candidate sequences that contains the correctsequence of hidden states with probability at least 1−ε. We presentexperimental results that confirm this holds in practice. We compare ourmethod with the standard approach (i.e.: the use of Maximum Likelihoodand the List–Viterbi algorithm), which suffers from violations tothe assumed distribution. We discuss advantages and limitations of ourmethod, and suggest future directions.

AB - We consider the problem of training a Hidden Markov Model(HMM) from fully observable data and predicting the hidden states of anobserved sequence. Our attention is focused to applications that requirea list of potential sequences as a prediction. We propose a novel methodbased on Conformal Prediction (CP) that, for an arbitrary confidencelevel 1 − ε, produces a list of candidate sequences that contains the correctsequence of hidden states with probability at least 1−ε. We presentexperimental results that confirm this holds in practice. We compare ourmethod with the standard approach (i.e.: the use of Maximum Likelihoodand the List–Viterbi algorithm), which suffers from violations tothe assumed distribution. We discuss advantages and limitations of ourmethod, and suggest future directions.

U2 - 10.1007/978-3-319-33395-3_10

DO - 10.1007/978-3-319-33395-3_10

M3 - Conference contribution

SN - 978-3-319-33394-6

VL - 9653

T3 - Lecture Notes in Computer Science

SP - 128

EP - 144

BT - Conformal and Probabilistic Prediction with Applications

PB - Springer

ER -