Language model combination and adaptation using weighted finite state transducers

In speech recognition systems language model (LMs) are often constructed by training and combining multiple n-gram models. They can be either used to represent different genres or tasks found in diverse text sources, or capture stochastic properties of different linguistic symbol sequences, for example, syllables and words. Unsupervised LM adaptation may also be used to further improve robustness to varying styles or tasks. When using these techniques, extensive software changes are often required. In this paper an alternative and more general approach based on weighted finite state transducers (WFSTs) is investigated for LM combination and adaptation. As it is entirely based on well-defined WFST operations, minimum change to decoding tools is needed. A wide range of LM combination configurations can be flexibly supported. An efficient on-the-fly WFST decoding algorithm is also proposed. Significant error rate gains of 7.3% relative were obtained on a state-of-the-art broadcast audio recognition task using a history dependently adapted multi-level LM modelling both syllable and word sequences. ©2010 IEEE.

Fixed-lag blind equalization and sequence estimation in digital communications systems using sequential importance sampling

We present methods for fixed-lag smoothing using Sequential Importance sampling (SIS) on a discrete non-linear, non-Gaussian state space system with unknown parameters. Our particular application is in the field of digital communication systems. Each input data point is taken from a finite set of symbols. We represent transmission media as a fixed filter with a finite impulse response (FIR), hence a discrete state-space system is formed. Conventional Markov chain Monte Carlo (MCMC) techniques such as the Gibbs sampler are unsuitable for this task because they can only perform processing on a batch of data. Data arrives sequentially, so it would seem sensible to process it in this way. In addition, many communication systems are interactive, so there is a maximum level of latency that can be tolerated before a symbol is decoded. We will demonstrate this method by simulation and compare its performance to existing techniques.

First real-time experimental demonstrations of 11.25Gb/s optical OFDMA PONs with adaptive dynamic bandwidth allocation.

End-to-end real-time experimental demonstrations are reported, for the first time, of aggregated 11.25Gb/s over 26.4km standard SMF, optical orthogonal frequency division multiple access (OOFDMA) PONs with adaptive dynamic bandwidth allocation (DBA). The demonstrated intensity-modulation and direct-detection (IMDD) OOFDMA PON system consists of two optical network units (ONUs), each of which employs a DFB-based directly modulated laser (DML) or a VCSEL-based DML for modulating upstream signals. Extensive experimental explorations of dynamic OOFDMA PON system properties are undertaken utilizing identified optimum DML operating conditions. It is shown that, for simultaneously achieving acceptable BERs for all upstream signals, the OOFDMA PON system has a >3dB dynamic ONU launch power variation range, and the BER performance of the system is insusceptible to any upstream symbol offsets slightly smaller than the adopted cyclic prefix. In addition, experimental results also indicate that, in addition to maximizing the aggregated system transmission capacity, adaptive DBA can also effectively reduce imperfections in transmission channel properties without affecting signal bit rates offered to individual ONUs.

Influencing factors of successful transitions towards product-service systems: A simulation approach

Product-Service Systems (PSS) are new business strategies moving and extending the product value towards its functional usage and related required services. From a theoretical point of view the PSS concept is known since a decade and many Authors reported reasonable possible success factors: higher profits over the entire life-cycle, diminished environmental burden, and localization of required services. Nevertheless the PSS promises remain quantitatively unproven relaying on a simple theory that involves a few constructs with some empirical grounding, but that is limited by weak conceptualization, few propositions, and/or rough underlying theoretical logic. A plausible interpretation to analyze the possible evolution of a PSS strategy could be considering it as a new business proposition competing on a traditional Product-Oriented (PO) market, assumed at its own equilibrium state at a given time. The analysis of the dynamics associated to a possible transition from a traditional PO to a PSS strategy allows investigating the main parameters and variables influencing an eventual successful adoption. This research is worthwhile because organizations undergoing fundamental PSS strategy are concerned about change and inertia key processes which, despite equilibrium theory and because of negative feedback loops, could undermine, economically, the return of their PSS proposition. In this paper Authors propose a qualitative System Dynamics (SD) approach by considering the PSS as a perturbation of an existing PO market featured by a set of known parameters. The proposed model incorporates several PSS factors able to influence the success of a PSS proposition under a set of given and justified assumptions, attempting to place this business strategy in a dynamic framework.

Mismatched shaping schemes for bit-interleaved coded modulation

We consider bit-interleaved coded modulation (BICM) schemes where, instead of the true bit or symbol probabilities and the constellation used at the transmitter, the decoder uses arbitrary probabilities or reference constellations. We study the corresponding low- and high- signal-to-noise-ratio regimes and show that even in the presence of this extra sources of mismatch, BICM has a negligible penalty with respect to coded modulation. © 2012 IEEE.

Efficient decoding with generative score-spaces using the expectation semiring

State-of-the-art speech recognisers are usually based on hidden Markov models (HMMs). They model a hidden symbol sequence with a Markov process, with the observations independent given that sequence. These assumptions yield efficient algorithms, but limit the power of the model. An alternative model that allows a wide range of features, including word- and phone-level features, is a log-linear model. To handle, for example, word-level variable-length features, the original feature vectors must be segmented into words. Thus, decoding must find the optimal combination of segmentation of the utterance into words and word sequence. Features must therefore be extracted for each possible segment of audio. For many types of features, this becomes slow. In this paper, long-span features are derived from the likelihoods of word HMMs. Derivatives of the log-likelihoods, which break the Markov assumption, are appended. Previously, decoding with this model took cubic time in the length of the sequence, and longer for higher-order derivatives. This paper shows how to decode in quadratic time. © 2013 IEEE.