Optimal Detection and Error Exponents for Hidden Semi-Markov Models

We study detection of random signals corrupted by noise that over time switch their values (states) between a finite set of possible values, where the switchings occur at unknown points in time. We model such signals as hidden semi-Markov signals, which generalize classical Markov chains by introducing explicit (possibly nongeometric) distribution for the time spent in each state. Assuming two possible signal states and Gaussian noise, we derive optimal likelihood ratio test and show that it has a computationally tractable form of a matrix product, with the number of matrices involved in the product being the number of process observations. The product matrices are independent and identically distributed, constructed by a simple measurement modulation of the sparse semi-Markov model transition matrix that we define in the paper. Using this result, we show that the Neyman–Pearson error exponent is equal to the top Lyapunov exponent for the corresponding random matrices. Using theory of large deviations, we derive a lower bound on the error exponent. Finally, we show that this bound is tight by means of numerical simulations.

[1]  Vladimir Stankovic,et al.  Optimal detection and error exponents for hidden multi-state processes via random duration model approach , 2017, ArXiv.

[2]  H. Vincent Poor,et al.  Neyman-pearson detection of gauss-Markov signals in noise: closed-form error exponentand properties , 2005, IEEE Transactions on Information Theory.

[3]  J. Doob.  Stochastic processes , 1953 .

[4]  Catherine Trottier,et al.  Markov and Semi‐Markov Switching Linear Mixed Models Used to Identify Forest Tree Growth Components , 2010, Biometrics.

[5]  Alfred O. Hero,et al.  Near-optimal signal detection for finite-state Markov signals with application to magnetic resonance force microscopy , 2006, IEEE Transactions on Signal Processing.

[6]  Jeffrey R. Russell,et al.  A Discrete-State Continuous-Time Model of Financial Transactions Prices and Times , 2005 .

[7]  Jing Liao,et al.  Non-intrusive appliance load monitoring using low-resolution smart meter data , 2014, 2014 IEEE International Conference on Smart Grid Communications (SmartGridComm).

[8]  Charles A. Sutton,et al.  Signal Aggregate Constraints in Additive Factorial HMMs, with Application to Energy Disaggregation , 2014, NIPS.

[9]  B. Lautrup,et al.  Products of random matrices. , 2002, Physical review. E, Statistical, nonlinear, and soft matter physics.

[10]  Jing Liao,et al.  Non-Intrusive Load Disaggregation Using Graph Signal Processing , 2018, IEEE Transactions on Smart Grid.

[11]  Ivett Lilian Flores.  DIRECT CALCULATION OF K-GENERALIZED FIBONACCI NUMBERS , 1967 .

[12]  Harald Luschgy,et al.  Asymptotic Behavior of Neyman-Pearson Tests for Autoregressive Processes , 1994 .

[13]  R. R. Bahadur.  Some Limit Theorems in Statistics , 1987 .

[14]  Stephen P. Boyd,et al.  Convex Optimization , 2004, Algorithms and Theory of Computation Handbook.

[15]  S. Natarajan,et al.  Large deviations, hypotheses testing, and source coding for finite Markov chains , 1985, IEEE Trans. Inf. Theory.

[16]  L. B. Boza.  Asymptotically Optimal Tests for Finite Markov Chains , 1971 .

[17]  V. Stanković,et al.  An electrical load measurements dataset of United Kingdom households from a two-year longitudinal study , 2017, Scientific Data.

[18]  Sheldon M. Ross,et al.  Stochastic Processes , 2018, Gauge Integral Structures for Stochastic Calculus and Quantum Electrodynamics.

[19]  Thomas M. Cover,et al.  Elements of information theory (2. ed.) , 2006 .

[20]  Mario Bergés,et al.  The Neural Energy Decoder : Energy Disaggregation by Combining Binary Subcomponents , 2016 .

[21]  Shunzheng Yu,et al.  Hidden semi-Markov models , 2010, Artif. Intell..

[22]  Po-Ning Chen.  General formulas for the Neyman-Pearson type-II error exponent subject to fixed and exponential type-I error bounds , 1996, IEEE Trans. Inf. Theory.

[23]  Christopher Uggen,et al.  Work as a Turning Point in the Life Course of Criminals: A Duration Model of Age, Employment, and Recidivism , 2000, American Sociological Review.

[24]  H. Vincent Poor,et al.  Neyman-Pearson Detection of Gauss-Markov Signals in Noise: Closed-Form Error Exponent and Properties , 2005, ISIT.

[25]  Yue M. Lu,et al.  Optimal Detection of Random Walks on Graphs: Performance Analysis via Statistical Physics , 2015, ArXiv.

[26]  Manish Marwah,et al.  Unsupervised Disaggregation of Low Frequency Power Measurements , 2011, SDM.

[27]  Jing Liao,et al.  Low-complexity energy disaggregation using appliance load modelling , 2016 .

[28]  H. Chernoff.  A Measure of Asymptotic Efficiency for Tests of a Hypothesis Based on the sum of Observations , 1952 .

[29]  Alfred O. Hero,et al.  Detection Of a Random Walk Signal in the Regime of Low Signal to Noise Ratio and Long Observation Time , 2006, 2006 IEEE International Conference on Acoustics Speech and Signal Processing Proceedings.

[30]  Daniel Kelly,et al.  Disaggregation of domestic smart meter energy data , 2016 .

[31]  L. H. Koopmans.  Asymptotic Rate of Discrimination for Markov Processes , 1960 .

[32]  Karol Vasek.  On the error exponent for ergodic Markov source , 1980, Kybernetika.

[33]  Silvia Chiappa,et al.  Explicit-Duration Markov Switching Models , 2014, Found. Trends Mach. Learn..

[34]  José M. F. Moura,et al.  Detecting Random Walks on Graphs With Heterogeneous Sensors , 2017, IEEE Transactions on Information Theory.

[35]  Tommi S. Jaakkola,et al.  Approximate Inference in Additive Factorial HMMs with Application to Energy Disaggregation , 2012, AISTATS.

[36]  Alex Rogers,et al.  Non-Intrusive Load Monitoring Using Prior Models of General Appliance Types , 2012, AAAI.

[37]  Dominik Egarter,et al.  Complexity of Power Draws for Load Disaggregation , 2015, ArXiv.

[38]  A. Rukhin,et al.  Adaptive tests for stochastic processes in the ergodic case , 1993 .

[39]  Vladimir Stankovic,et al.  On a Training-Less Solution for Non-Intrusive Appliance Load Monitoring Using Graph Signal Processing , 2016, IEEE Access.

[40]  John N. Tsitsiklis,et al.  The Lyapunov exponent and joint spectral radius of pairs of matrices are hard—when not impossible—to compute and to approximate , 1997, Math. Control. Signals Syst..

[41]  Giuseppe Longo,et al.  The error exponent for the noiseless encoding of finite ergodic Markov sources , 1981, IEEE Trans. Inf. Theory.