The Power of Amnesia: Learning Probabilistic Automata with Variable Memory Length

By Dana Ron, Yoram Singer, Naftali Tishby. Machine-mediated learning, 2005

530 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Molecular Biology

View Paper: https://www.semanticscholar.org/paper/9e5e82a07998ce355b65fe5d43b5c3138fd767d9

How researchers see this paper

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

In Artificial Intelligence

1. Yoshua Bengio
2. Sergey Levine
3. Marc G. Bellemare
4. Charu C. Aggarwal
5. Pieter Abbeel
6. Joelle Pineau
7. Yun Raymond Fu
8. Ruslan Salakhutdinov

In Molecular Biology

1. Cosma Rohilla Shalizi
2. Kristina Lisa Shalizi
3. Yoshua Bengio
4. Ran El-Yaniv
5. Joel Veness
6. Arindam Banerjee
7. Martin Mächler
8. Peter Buhlmann

Citations in the corpus

• Beyond Word Frequency: Bursts, Lulls, and Scaling in the Temporal Distributions of Words (2009, background): “The model we develop to explain these observations shares the generative spirit of local (n-gram) and weakly non-local models of text classification and generation [29, 30, 31].”
• Blind Construction of Optimal Nonlinear Recursive Predictors for Discrete Sequences (2004, methodology): “Suffix-tree methods can induce POMDPs from data [23, 27, 28], and we believe CSSR can be adapted to reconstruct POMDPs.”
• Modeling Complex Temporal Composition of Actionlets for Activity Prediction (2012, methodology): “Here, we model complex human activity as a Probabilistic Suffix Tree (PST) [21] which implements the single best L -bounded VMM (VMMs of degree L or less) in a fast and efficient way.”
• The Continuator: Musical Interaction With Style (2003, methodology): “In this framework, the state of the Markov model are not simply the items of input sequences, as other, hidden state are inferred, precisely to represent state regions, and eventually cope with inexact string inputs.”
• Anomaly Detection for Discrete Sequences: A Survey (2012, methodology): “Markovian techniques utilize the short memory property of sequences, which is observed across different domains [28].”
• Skip Context Tree Switching (2014, methodology): “A powerful algorithm for this setting is Context Tree Weighting (CTW, Willems et al., 1995), which efficiently performs Bayesian model averaging over a class of prediction suffix trees (Ron et al., 1996).”
• Markovian Models for Sequential Data (2004, methodology): “A Variable Length Markov Model [12] is a probability model over strings in which the state variable is not hidden: its value is a deterministic function of the past observation sequence.”
• Variable Length Markov Chains: Methodology, Computing, and Software (2004, methodology): “…1983); related models and methods, from information theory and machine learning, are context tree weighting (Willems, Shtarkov & Tjalkens 1996) or probabilistic suffix trees (Ron, Singer & Tishby 1996). vlmc ’s have been recently discussed from a more statistical view (B¨uhlmann & Wyner 1999).”
• On Prediction Using Variable Order Markov Models (2004, methodology): “The Probabilistic Suffix Tree (pst) prediction algorithm (Ron et al., 1996) attempts to construct the single “best” D-bounded VMM according to the training sequence.”
• Prediction of Human Activity by Discovering Temporal Sequence Patterns (2014, methodology): “Here, we model complex human activity as a probabilistic suffix tree [44] which imple-”