Applications of stochastic context-free grammars using the Inside-Outside algorithm

This paper describes two applications in speech recognition of the use of stochastic context-free grammars (SCFGs) trained automatically via the Inside-Outside Algorithm. First, SCFGs are used to model VQ encoded speech for isolated word recognition and are compared directly to HMMs used for the same task. It is shown that SCFGs can model this low-level VQ data accurately and that a regular grammar based pre-training algorithm is effective both for reducing training time and obtaining robust solutions. Second, an SCFG is inferred from a transcription of the speech used to train a phoneme-based recognizer in an attempt to model phonotactic constraints. When used as a language model, this SCFG gives improved performance over a comparable regular grammar or bigram. © 1991.

Low power all-optical switch in a passive InGaAsP multiple quantum well waveguide

We demonstrate the use of resonant bandfilling nonlinearity in an InGaAsP/InGaAsP Multiple Quantum Well (MQW) waveguide due to photogenerated carriers to obtain switching at pulse powers, which can readily be obtained from an erbium amplified diode laser source. In order to produce gating a polarisation rotation gate was used, which relies on an asymmetry in the nonlinear refraction on the principle axes of the waveguide.

HVIC switch for high-frequency power converters

A high voltage integrated circuit (HVIC) switch designed as a building block for power converters operating up to 13.56 MHz from off-line voltages is presented. A CMOS-compatible, 500 V power device process is used to integrate control circuitry with a high-speed MOS gate driver and high voltage lateral power MOSFET. Fabrication of the HVIC switches has proceeded in two stages. The first batch of devices showed switching times of less than 5 ns for the power switch and good high frequency performance of a level-shifter for driving half bridge converters. In the second phase, a switch that monolithically integrates all the elements required to form a complete high-frequency converter has been designed.

Three band-gap QW intermixing in InP/InGaAs/InGaAsP system for monolithically integrated optical switch

Quantum well intermixing is a key technique for photonic integration. The intermixing of InP/InGaAs/InGaAsP material involving the deposition of a layer of sputtered SiO2 on the semiconductor surface, followed by thermal annealing has allowed good control of the intermixing process and has been used to fabricate extended cavity lasers. This will be used for optimization of the performance of optical switches consisting of passive components, modulators and amplifiers.

Process development on the monolithic fabrication of an ultra-compact 4×4 optical switch matrix on InP/InGaAsP material

We have fabricated an ultra-compact 4×4 optical matrix on InP/InGaAsP material. 1×4 MMI couplers and TIR mirrors are employed to produce a compact 1×2 mm2 device. A CH4/H2/O2 RIE dry etch process has been used to realize two-level dry etching: deep-etch for both the MMI couplers and the mirrors and shallow-etch for the rest of the routing waveguides. It was found that a metal/dielectric bilayer mask is essential for multi-dry-etch processes and high profile verticality. We have found a Ti intermediate mask for the deep-etch process which is removable by SF6 dry-etch before the following shallow process. Dry-etch removal of the intermediate mask is necessary to protect the deep-etched mirror sidewall.

All-optical switching in a vertical coupler space switch employing photocarrier-induced nonlinearity

A novel compact integrated nonlinear optical switch is demonstrated. Using a high-power picosecond pulse of 5-ps pulsewidth and 250-MHz repetition rate, all-optical switching with a contrast ratio of 23 dB has been achieved using an in-fiber input power < 14 dBm (100 pJ/pulse). The switch speed depends on the carrier sweep-out time, which can be reduced to the 10 ps range by either applying a reverse bias or by introduction of carrier recombination centers in the active layer.

Decision tree-based context clustering based on cross validation and hierarchical priors

The standard, ad-hoc stopping criteria used in decision tree-based context clustering are known to be sub-optimal and require parameters to be tuned. This paper proposes a new approach for decision tree-based context clustering based on cross validation and hierarchical priors. Combination of cross validation and hierarchical priors within decision tree-based context clustering offers better model selection and more robust parameter estimation than conventional approaches, with no tuning parameters. Experimental results on HMM-based speech synthesis show that the proposed approach achieved significant improvements in naturalness of synthesized speech over the conventional approaches. © 2011 IEEE.

The need for winter in the switch to flowering.

Vernalization is the process whereby the floral transition is promoted through exposure of plants to long periods of cold temperature or winter. A requirement for vernalization aligns flowering with the seasons to ensure that their reproductive phase occurs in favorable conditions. The mitotic stability of vernalization, suggestive of an epigenetic mechanism, has intrigued researchers for many years. Genetic analysis of the vernalization requirement in Arabidopsis has identified key floral repressor genes, FRI and FLC. The action of these floral repressors is antagonized by vernalization and the activity of a set of genes grouped into the autonomous floral pathway. Analysis of the vernalization pathway has defined a series of epigenetic regulators crucial for "cellular-memory" of the cold signal, whereas the autonomous pathway appears to function in part through posttranscriptional mechanisms. The mechanism of the vernalization requirement, which is now being explored in a range of plant species, should uncover the evolutionary origins of this key agronomic trait.

A single-rate context-dependent learning process underlies rapid adaptation to familiar object dynamics.

Motor learning has been extensively studied using dynamic (force-field) perturbations. These induce movement errors that result in adaptive changes to the motor commands. Several state-space models have been developed to explain how trial-by-trial errors drive the progressive adaptation observed in such studies. These models have been applied to adaptation involving novel dynamics, which typically occurs over tens to hundreds of trials, and which appears to be mediated by a dual-rate adaptation process. In contrast, when manipulating objects with familiar dynamics, subjects adapt rapidly within a few trials. Here, we apply state-space models to familiar dynamics, asking whether adaptation is mediated by a single-rate or dual-rate process. Previously, we reported a task in which subjects rotate an object with known dynamics. By presenting the object at different visual orientations, adaptation was shown to be context-specific, with limited generalization to novel orientations. Here we show that a multiple-context state-space model, with a generalization function tuned to visual object orientation, can reproduce the time-course of adaptation and de-adaptation as well as the observed context-dependent behavior. In contrast to the dual-rate process associated with novel dynamics, we show that a single-rate process mediates adaptation to familiar object dynamics. The model predicts that during exposure to the object across multiple orientations, there will be a degree of independence for adaptation and de-adaptation within each context, and that the states associated with all contexts will slowly de-adapt during exposure in one particular context. We confirm these predictions in two new experiments. Results of the current study thus highlight similarities and differences in the processes engaged during exposure to novel versus familiar dynamics. In both cases, adaptation is mediated by multiple context-specific representations. In the case of familiar object dynamics, however, the representations can be engaged based on visual context, and are updated by a single-rate process.