Publications

Refine Results

(Filters Applied) Clear All

Efficient speech translation through confusion network decoding

Published in:
IEEE Trans. Audio Speech Lang. Proc., Vol. 16, No. 8, November 2008, pp. 1696-1705.

Summary

This paper describes advances in the use of confusion networks as interface between automatic speech recognition and machine translation. In particular, it presents a decoding algorithm for confusion networks which results as an extension of a state-of-the-art phrase-based text translation decoder. The confusion network decoder significantly improves both in efficiency and performance over previous work along this direction, and outperforms the background text translation system. Experimental results in terms of translation accuracy and decoding efficiency are reported for the task of translating plenary speeches of the European Parliament from Spanish to English and from English to Spanish.
READ LESS

Summary

This paper describes advances in the use of confusion networks as interface between automatic speech recognition and machine translation. In particular, it presents a decoding algorithm for confusion networks which results as an extension of a state-of-the-art phrase-based text translation decoder. The confusion network decoder significantly improves both in efficiency...

READ MORE

Detection of condensed-phase explosives via laser-induced vaporization, photodissociation, and resonant excitation

Published in:
Appl. Opt., Vol. 47, No. 31, 1 November 2008, pp. 5767-5776.

Summary

We investigate the remote detection of explosives via a technique that vaporizes and photodissociates the condensed-phase material and detects the resulting vibrationally excited NO fragments via laser-induced fluorescence. The technique utilizes a single 7 ns pulse of a tunable laser near 236:2nm to perform these multiple processes. The resulting blue-shifted fluorescence (226 nm) is detected using a photomultiplier and narrowband filter that strongly block the scatter of the pump laser off the solid media while passing the shorter wavelength photons. Various nitro-bearing compounds, including 2,6-dinitrotoluene (DNT), 2,4,6-trinitrotoluene (TNT), pentaerythritol tetranitrate (PETN), and hexahydro-1,3,5- trinitro-1,3,5-triazine (RDX) were detected with a signal-to-noise of 25 dB. The effects of laser fluence, wavelength, and sample morphology were examined.
READ LESS

Summary

We investigate the remote detection of explosives via a technique that vaporizes and photodissociates the condensed-phase material and detects the resulting vibrationally excited NO fragments via laser-induced fluorescence. The technique utilizes a single 7 ns pulse of a tunable laser near 236:2nm to perform these multiple processes. The resulting blue-shifted...

READ MORE

A polyphase nonlinear equalization architecture and semi-blind identification method

Published in:
42th Asilomar Conf. on Signals, Systems, and Computers, 27 October 2008, pp. 593-597.

Summary

In this paper, we present an architecture and semiblind identification method for a polyphase nonlinear equalizer (pNLEQ). Such an equalizer is useful for extending the dynamic range of time-interleaved analog-to-digital converters (ADCs). Our proposed architecture is a polyphase extension to other architectures that partition the Volterra kernel into small nonlinear filters with relatively low computational complexity. Our semi-blind identification technique addresses important practical concerns in the equalizer identification process. We describe our architecture and demonstrate its performance with measured results when applied to a National Semiconductor ADC081000.
READ LESS

Summary

In this paper, we present an architecture and semiblind identification method for a polyphase nonlinear equalizer (pNLEQ). Such an equalizer is useful for extending the dynamic range of time-interleaved analog-to-digital converters (ADCs). Our proposed architecture is a polyphase extension to other architectures that partition the Volterra kernel into small nonlinear...

READ MORE

The cube coefficient subspace architecture for nonlinear digital predistortion

Published in:
42th Asilomar Conf. on Signals, Systems, and Computers, 27 October 2008, pp. 1857-1861.

Summary

In this paper, we present the cube coefficient subspace (CCS) architecture for linearizing power amplifiers (PAs), which divides the overparametrized Volterra kernel into small, computationally efficient subkernels spanning only the portions of the full multidimensional coefficient space with the greatest impact on linearization. Using measured results from a Q-Band solid state PA, we demonstrate that the CCS predistorter architecture achieves better linearization performance than state-of-the-art memory polynomials and generalized memory polynomials.
READ LESS

Summary

In this paper, we present the cube coefficient subspace (CCS) architecture for linearizing power amplifiers (PAs), which divides the overparametrized Volterra kernel into small, computationally efficient subkernels spanning only the portions of the full multidimensional coefficient space with the greatest impact on linearization. Using measured results from a Q-Band solid...

READ MORE

Correlated encounter model for cooperative aircraft in the National Airspace System, version 1.0

Published in:
MIT Lincoln Laboratory Report ATC-344

Summary

This document describes a new cooperative aircraft encounter model for the National Airspace System (NAS). The model is used to generate random close encounters between transponder-equipped (cooperative) aircraft in fast-time Monte Carlo simulations to evaluate collision avoidance system concepts. An extensive set of radar data from across the United States, including more than 120 sensors and collected over a period of nine months, was used to build the statistical relationships in the model to ensure that the encounters that are generated are representative of actual events in the airspace.
READ LESS

Summary

This document describes a new cooperative aircraft encounter model for the National Airspace System (NAS). The model is used to generate random close encounters between transponder-equipped (cooperative) aircraft in fast-time Monte Carlo simulations to evaluate collision avoidance system concepts. An extensive set of radar data from across the United States...

READ MORE

Characterization of a three-dimensional SOI integrated-circuit technology

Published in:
2008 IEEE Int. SOI Conf. Proc., 6 October 2008, pp. 109-110.

Summary

At Lincoln Laboratory, we have established a three dimensional (3D) integrated circuit (IC) technology that has been developed and demonstrated over eight designs, bonding two or three active circuit layers or tiers to form monolithically integrated 3D circuits. This technology has been used to successfully demonstrate a large-area 8 x 8 mm2 high-3D-via-count 1024 x 1024 visible image, a 64 x 64 laser radar focal plane based on single-photon-sensitive avalanche photodiodes, and a 10Gb/s/pin low power interconnect for 3DICs. 3DIC technology in our most recently completed 3D multiproject (3DM2) run includes three active fully-depleted-SOI (FDSOI) circuit tiers, eleven interconnect-metal layers, and dense unrestricted 3D vias interconnecting stacked circuit layers, as shown in Figure 1. While we continue our efforts to scale our 3DIC technology and increase 3D via density, we are also working to improve our understanding of 3D integration impact on transistor and process monitor circuits. In this paper, we describe our process and test results after single tier circuit fabrication as well as after three-tier integration, determine impact of 3D vias on ring oscillator performance, and demonstrate functionality of single and multi-tier circuits of varying complexity.
READ LESS

Summary

At Lincoln Laboratory, we have established a three dimensional (3D) integrated circuit (IC) technology that has been developed and demonstrated over eight designs, bonding two or three active circuit layers or tiers to form monolithically integrated 3D circuits. This technology has been used to successfully demonstrate a large-area 8 x...

READ MORE

Operational usage of the Route Availability Planning Tool during the 2007 convective weather season : executive summary

Published in:
MIT Lincoln Laboratory Report ATC-339-1

Summary

The Route Availability Planning Tool (RAPT) is an integrated weather/air traffic management decision support tool that has been designed to help traffic managers better anticipate weather impacts on jet routes and thus improve NY departure route usage efficiency. A field study was conducted in 2007 to evaluate RAPT technical performance at forecasting route blockage, to assess RAPT operational use during adverse weather, and to evaluate RAPT benefits. The operational test found that RAPT guidance was operationally sound and timely in many circumstances. RAPT applications included increased departure route throughput, more efficient reroute planning, and more timely decision coordination. Estimated annual NY departure delay savings attributed to RAPT in 2007 totaled 2,300 hours, with a cost savings of $7.5 M. The RAPT field study also sought to develop a better understanding of NY traffic flow decision-making during convective weather impacts since the RAPT benefi ts in 2007 were significantly limited by a number of factors other than direct weather impacts. Observations were made of the multi-facility departure management decision chain, the traffic management concerns and responsibilities at specific FAA facilities, and the procedures and pitfalls of the current process for capturing and disseminating key information such as route/fix availability and restrictions.
READ LESS

Summary

The Route Availability Planning Tool (RAPT) is an integrated weather/air traffic management decision support tool that has been designed to help traffic managers better anticipate weather impacts on jet routes and thus improve NY departure route usage efficiency. A field study was conducted in 2007 to evaluate RAPT technical performance...

READ MORE

Operational usage of the Route Availability Planning Tool during the 2007 convective weather season

Published in:
MIT Lincoln Laboratory Report ATC-339

Summary

The Route Availability Planning Tool (RAPT) is an integrated weather/air traffic management decision support tool that has been designed to help traffic managers better anticipate weather impacts on jet routes and thus improve NY departure route usage efficiency. A field study was conducted in 2007 to evaluate RAPT technical performance at forecasting route blockage, to assess RAPT operational use during adverse weather, and to evaluate RAPT benefits. The operational test found that RAPT guidance was operationally sound and timely in many circumstances. RAPT applications included increased departure route throughput, more efficient reroute planning, and more timely decision coordination. Estimated annual NY departure delay savings attributed to RAPT in 2007 totaled 2,300 hours, with a cost savings of $7.5 M. The RAPT field study also sought to develop a better understanding of NY traffic flow decision-making during convective weather impacts since the RAPT benefits in 2007 were significantly limited by a number of factors other than direct weather impacts. Observations were made of the multi-facility departure management decision chain, the traffic management concerns and responsibilities at specific FAA facilities, and the procedures and pitfalls of the current process for capturing and disseminating key information such as route/fix availability and restrictions.
READ LESS

Summary

The Route Availability Planning Tool (RAPT) is an integrated weather/air traffic management decision support tool that has been designed to help traffic managers better anticipate weather impacts on jet routes and thus improve NY departure route usage efficiency. A field study was conducted in 2007 to evaluate RAPT technical performance...

READ MORE

InP-based single-photon detector arrays with asynchronous readout integrated circuits

Summary

We have developed and demonstrated a highduty- cycle asynchronous InGaAsP-based photon counting detector system with near-ideal Poisson response, roomtemperature operation, and nanosecond timing resolution for near-infrared applications. The detector is based on an array of Geiger-mode avalanche photodiodes coupled to a custom integrated circuit that provides for lossless readout via an asynchronous, nongated architecture. We present results showing Poisson response for incident photon flux rates up to 10 million photons per second and multiple photons per 3-ns timing bin.
READ LESS

Summary

We have developed and demonstrated a highduty- cycle asynchronous InGaAsP-based photon counting detector system with near-ideal Poisson response, roomtemperature operation, and nanosecond timing resolution for near-infrared applications. The detector is based on an array of Geiger-mode avalanche photodiodes coupled to a custom integrated circuit that provides for lossless readout via...

READ MORE

Language, dialect, and speaker recognition using Gaussian mixture models on the cell processor

Published in:
Twelfth Annual High Performance Embedded Computing Workshop, HPEC 2008, 23-25 September 2008.

Summary

Automatic recognition systems are commonly used in speech processing to classify observed utterances by the speaker's identity, dialect, and language. These problems often require high processing throughput, especially in applications involving multiple concurrent incoming speech streams, such as in datacenter-level processing. Recent advances in processor technology allow multiple processors to reside within the same chip, allowing high performance per watt. Currently the Cell Broadband Engine has the leading performance-per-watt specifications in its class. Each Cell processor consists of a PowerPC Processing Element (PPE) working together with eight Synergistic Processing Elements (SPE). The SPEs have 256KB of memory (local store), which is used for storing both program and data. This paper addresses the implementation of language, dialect, and speaker recognition on the Cell architecture. Classically, the problem of performing speech-domain recognition has been approached as embarrassingly parallel, with each utterance being processed in parallel to the others. As we will discuss, efficient processing on the Cell requires a different approach, whereby computation and data for each utterance are subdivided to be handled by separate processors. We present a computational model for automatic recognition on the Cell processor that takes advantage of its architecture, while mitigating its limitations. Using the proposed design, we predict a system able to concurrently score over 220 real-time speech streams on a single Cell.
READ LESS

Summary

Automatic recognition systems are commonly used in speech processing to classify observed utterances by the speaker's identity, dialect, and language. These problems often require high processing throughput, especially in applications involving multiple concurrent incoming speech streams, such as in datacenter-level processing. Recent advances in processor technology allow multiple processors to...

READ MORE