An Online Expectation-Maximization Algorithm for Changepoint Models

Changepoint models are widely used to model the heterogeneity of sequential data. We present a novel sequential Monte Carlo (SMC) online Expectation-Maximization (EM) algorithm for estimating the static parameters of such models. The SMC online EM algorithm has a cost per time which is linear in the number of particles and could be particularly important when the data is representable as a long sequence of observations, since it drastically reduces the computational requirements for implementation. We present an asymptotic analysis for the stability of the SMC estimates used in the online EM algorithm and demonstrate the performance of this scheme using both simulated and real data originating from DNA analysis.

A reason for unreason: returns-based beliefs in game theory

Players cooperate in experiments more than game theory would predict. We introduce the ‘returns-based beliefs’ approach: the expected returns of a particular strategy in proportion to total expected returns of all strategies. Using a decision analytic solution concept, Luce’s (1959) probabilistic choice model, and ‘hyperpriors’ for ambiguity in players’ cooperability, our approach explains empirical observations in various classes of games including the Prisoner’s and Traveler’s Dilemmas. Testing the closeness of fit of our model on Selten and Chmura (2008) data for completely mixed 2 × 2 games shows that with loss aversion, returns-based beliefs explain the data better than other equilibrium concepts.

Forecasting high-frequency futures returns using online langevin dynamics

Forecasting the returns of assets at high frequency is the key challenge for high-frequency algorithmic trading strategies. In this paper, we propose a jump-diffusion model for asset price movements that models price and its trend and allows a momentum strategy to be developed. Conditional on jump times, we derive closed-form transition densities for this model. We show how this allows us to extract a trend from high-frequency finance data by using a Rao-Blackwellized variable rate particle filter to filter incoming price data. Our results show that even in the presence of transaction costs our algorithm can achieve a Sharpe ratio above 1 when applied across a portfolio of 75 futures contracts at high frequency. © 2011 IEEE.

Integrated online speaker clustering and adaptation

For many applications, it is necessary to produce speech transcriptions in a causal fashion. To produce high quality transcripts, speaker adaptation is often used. This requires online speaker clustering and incremental adaptation techniques to be developed. This paper presents an integrated approach to online speaker clustering and adaptation which allows efficient clustering of speakers using the same accumulated statistics that are normally used for adaptation. Using a consistent criterion for both clustering and adaptation should yield gains for both stages. The proposed approach is evaluated on a meetings transcription task using audio from multiple distant microphones. Consistent gains over standard clustering and adaptation were obtained. Copyright © 2011 ISCA.

An online expectation-maximisation algorithm for nonnegative matrix factorisation models

In this paper we formulate the nonnegative matrix factorisation (NMF) problem as a maximum likelihood estimation problem for hidden Markov models and propose online expectation-maximisation (EM) algorithms to estimate the NMF and the other unknown static parameters. We also propose a sequential Monte Carlo approximation of our online EM algorithm. We show the performance of the proposed method with two numerical examples. © 2012 IFAC.

IPLR: An online resource for Greek word-level and sublexical information

We present a new online psycholinguistic resource for Greek based on analyses of written corpora combined with text processing technologies developed at the Institute for Language & Speech Processing (ILSP), Greece. The "ILSP PsychoLinguistic Resource" (IPLR) is a freely accessible service via a dedicated web page, at http://speech.ilsp.gr/iplr. IPLR provides analyses of user-submitted letter strings (words and nonwords) as well as frequency tables for important units and conditions such as syllables, bigrams, and neighbors, calculated over two word lists based on printed text corpora and their phonetic transcription. Online tools allow retrieval of words matching user-specified orthographic or phonetic patterns. All results and processing code (in the Python programming language) are freely available for noncommercial educational or research use. © 2010 Springer Science+Business Media B.V.

Linear quadratic game and non-cooperative predictive methods for potential application to modelling driver-AFS interactive steering control

This paper is concerned with the modelling of strategic interactions between the human driver and the vehicle active front steering (AFS) controller in a path-following task where the two controllers hold different target paths. The work is aimed at extending the use of mathematical models in representing driver steering behaviour in complicated driving situations. Two game theoretic approaches, namely linear quadratic game and non-cooperative model predictive control (non-cooperative MPC), are used for developing the driver-AFS interactive steering control model. For each approach, the open-loop Nash steering control solution is derived; the influences of the path-following weights, preview and control horizons, driver time delay and arm neuromuscular system (NMS) dynamics are investigated, and the CPU time consumed is recorded. It is found that the two approaches give identical time histories as well as control gains, while the non-cooperative MPC method uses much less CPU time. Specifically, it is observed that the introduction of weight on the integral of vehicle lateral displacement error helps to eliminate the steady-state path-following error; the increase in preview horizon and NMS natural frequency and the decline in time delay and NMS damping ratio improve the path-following accuracy. © 2013 Copyright Taylor and Francis Group, LLC.

How can online schedules improve communication and estimation tradeoff?

We consider remote state estimation and investigate the tradeoff between the sensor-to-estimator communication rate and the remote estimation quality. It is well known that if the communication rate is one, e.g., the sensor communicates with the remote estimator at each time, then the remote estimation quality is the best. It degrades when the communication rate drops. We present one optimal offline schedule and two online schedules and show that the two online schedules provide better tradeoff between the communication rate and the estimation quality than the optimal offline schedule. Simulation examples demonstrate that significant communication savings can be achieved under the two online schedules which only introduce small increment of the estimation errors. © 1991-2012 IEEE.

Simulations to design an online motion compensation system for scanned particle beams

Respiration-induced target motion is a major problem in intensity-modulated radiation therapy. Beam segments are delivered serially to form the total dose distribution. In the presence of motion, the spatial relation between dose deposition from different segments will be lost. Usually, this results in over-and underdosage. Besides such interplay effects between target motion and dynamic beam delivery as known from photon therapy, changes in internal density have an impact on delivered dose for intensity-modulated charged particle therapy. In this study, we have analysed interplay effects between raster scanned carbon ion beams and target motion. Furthermore, the potential of an online motion strategy was assessed in several simulations. An extended version of the clinical treatment planning software was used to calculate dose distributions to moving targets with and without motion compensation. For motion compensation, each individual ion pencil beam tracked the planned target position in the lateral aswell as longitudinal direction. Target translations and rotations, including changes in internal density, were simulated. Target motion simulating breathing resulted in severe degradation of delivered dose distributions. For example, for motion amplitudes of +/- 15 mm, only 47% of the target volume received 80% of the planned dose. Unpredictability of resulting dose distributions was demonstrated by varying motion parameters. On the other hand, motion compensation allowed for dose distributions for moving targets comparable to those for static targets. Even limited compensation precision (standard deviation similar to 2 mm), introduced to simulate possible limitations of real-time target tracking, resulted in less than 3% loss in dose homogeneity.

An online monitor ionization chamber used in particle therapy

The clinical trials of tumor therapy using heavy ions beam C-12 are now in progress at Institute of Modem Physics in Lanzhou. In order to achieve the precise radiotherapy with the high energy C-12 beam in active pencil beam scanning mode, we have developed an ionization chamber(IC) as an online monitor for beam intensity and also a dosimeter after calibration. Through the choosing of working gas and voltage, optimizing of the electrics and the read-out system, calibrating the linearity, the detector system provide us one of the simple and highly reliable way to monitoring the beam during the active pencil beam scanning treatments. The measurement results of this detector system show that it could work well under the condition of high energy C-12 beam in active pencil beam scanning mode.