Publications

Refine Results

(Filters Applied) Clear All

Using United States government language proficiency standards for MT evaluation

Published in:
Chapter 5.3.3 in Handbook of Natural Language Processing and Machine Translation, 2011, pp. 775-82.

Summary

The purpose of this section is to discuss a method of measuring the degree to which the essential meaning of the original text is communicated in the MT output. We view this test to be a measurement of the fundamental goal of MT; that is, to convey information accurately from one language to another. We conducted a series of experiments in which educated native readers of English responded to test questions about translated versions of texts originally written in Arabic and Chinese. We compared the results for those subjects using machine translations of the texts with those using professional reference translations. These comparisons serve as a baseline for determining the level of foreign language reading comprehension that can be achieved by a native English reader relying on machine translation technology. This also allows us to explore the relationship between the current, broadly accepted automatic measures of performance for machine translation and a test derived from the Defense Language Proficiency Test, which is used throughout the Defense Department for measuring foreign language proficiency. Our goal is to put MT system performance evaluation into terms that are meaningful to US government consumers of MT output.
READ LESS

Summary

The purpose of this section is to discuss a method of measuring the degree to which the essential meaning of the original text is communicated in the MT output. We view this test to be a measurement of the fundamental goal of MT; that is, to convey information accurately from...

READ MORE

MIT Lincoln Laboratory TCAS surveillance performance

Published in:
MIT Lincoln Laboratory Report ATC-370

Summary

The Traffic Alert and Collision Avoidance System (TCAS) Version 7 surveillance requirements were developed in the mid-1990s with the use of limited radar data. Recently, a more comprehensive radar data source has become available, enabling a thorough analysis of TCAS surveillance performance throughouth the National Airspace System (NAS). This paper characterizes six high traffic terminal environments over three months. A busy one hour period was selected from each location for density and equipage measurements. This paper then describes the use of a high fidelity simulation to characterize TCAS surveillance performance in the isx locations. Transponder utilization due to TCAS and TCAS surveillance range are compared with the design requirements, including interference limiting specifications. The effect of TCAS surveillance activity on Air Traffic Control (ATC) ground radar performance is also investigated. Results indicate that the surveillance algorithms perform as intended and that TCAS has a minimal impact on ground radar. Areas of concern are noted for future investigation.
READ LESS

Summary

The Traffic Alert and Collision Avoidance System (TCAS) Version 7 surveillance requirements were developed in the mid-1990s with the use of limited radar data. Recently, a more comprehensive radar data source has become available, enabling a thorough analysis of TCAS surveillance performance throughouth the National Airspace System (NAS). This paper...

READ MORE

Direct and latent modeling techniques for computing spoken document similarity

Published in:
SLT 2010, IEEE Workshop on Spoken Language Technology, 12-15 December 2010.

Summary

Document similarity measures are required for a variety of data organization and retrieval tasks including document clustering, document link detection, and query-by-example document retrieval. In this paper we examine existing and novel document similarity measures for use with spoken document collections processed with automatic speech recognition (ASR) technology. We compare direct vector space approaches using the cosine similarity measure applied to feature vectors constructed with various forms of term frequency inverse document frequency (TF-IDF) normalization against latent topic modeling approaches based on latent Dirichlet allocation (LDA). In document link detection experiments on the Fisher Corpus, we find that an approach that applies bagging to models derived from LDA substantially outperforms the direct vector space approach.
READ LESS

Summary

Document similarity measures are required for a variety of data organization and retrieval tasks including document clustering, document link detection, and query-by-example document retrieval. In this paper we examine existing and novel document similarity measures for use with spoken document collections processed with automatic speech recognition (ASR) technology. We compare...

READ MORE

Subgraph detection using eigenvector L1 norms

Published in:
23rd Int. Conf. on Neural Info. Process. Syst., NIPS, 6-9 December 2010, pp. 1633-41.

Summary

When working with network datasets, the theoretical framework of detection theory for Euclidean vector spaces no longer applies. Nevertheless, it is desirable to determine the detectability of small, anomalous graphs embedded into background networks with known statistical properties. Casting the problem of subgraph detection in a signal processing context, this article provides a framework and empirical results that elucidate a "detection theory" for graph-valued data. Its focus is the detection of anomalies in unweighted, undirected graphs through L1 properties of the eigenvectors of the graph's so-called modularity matrix. This metric is observed to have relatively low variance for certain categories of randomly-generated graphs, and to reveal the presence of an anomalous subgraph with reasonable reliability when the anomaly is not well-correlated with stronger portions of the background graph. An analysis of subgraphs in real network datasets confirms the efficacy of this approach.
READ LESS

Summary

When working with network datasets, the theoretical framework of detection theory for Euclidean vector spaces no longer applies. Nevertheless, it is desirable to determine the detectability of small, anomalous graphs embedded into background networks with known statistical properties. Casting the problem of subgraph detection in a signal processing context, this...

READ MORE

Analysis of ground-based radar low-altitude wind-shear detection in OEP terminal airspace for NextGen

Author:
Published in:
MIT Lincoln Laboratory Report ATC-375

Summary

To support the Next Generation Air Transportation System (NextGen), the Reduce Weather Impact Sensor RightSizing program is identifying and analyzing gaps in the current sensor network coverage relative to the Four-Dimensional Weather Data Cube Single Authoritative Source performance requirements. In this study, we look for shortfalls in low-altitude wind-shear sensing by ground-based radars and lidar in the NextGen super-density operations (SDO) terminal airspace. Specifically, 2D gridded wind-shear visibility (an upper bound to detection probability) data are generated for microbursts and gust fronts separately for different sensors, namely the Terminal Doppler Weather Radar, Next Generation Weather Radar, Airport Surveillance Radar-9 with Weather Systems Processor, and Doppler lidar.
READ LESS

Summary

To support the Next Generation Air Transportation System (NextGen), the Reduce Weather Impact Sensor RightSizing program is identifying and analyzing gaps in the current sensor network coverage relative to the Four-Dimensional Weather Data Cube Single Authoritative Source performance requirements. In this study, we look for shortfalls in low-altitude wind-shear sensing...

READ MORE

The MIT-LL/AFRL IWSLT-2010 MT system

Published in:
Proc. Int. Workshop on Spoken Language Translation, IWSLT, 2 December 2010.

Summary

This paper describes the MIT-LUAFRL statistical MT system and the improvements that were developed during the IWSLT 2010 evaluation campaign. As part of these efforts, we experimented with a number of extensions to the standard phrase-based model that improve performance on the Arabic and Turkish to English translation tasks. We also participated in the new French to English BTEC and English to French TALK tasks. We discuss the architecture of the MIT-LL/AFRL MT system, improvements over our 2008 system, and experiments we ran during the IWSLT-2010 evaluation. Specifically, we focus on 1) cross-domain translation using MAP adaptation, 2) Turkish morphological processing and translation, 3) improved Arabic morphology for MT preprocessing, and 4) system combination methods for machine translation.
READ LESS

Summary

This paper describes the MIT-LUAFRL statistical MT system and the improvements that were developed during the IWSLT 2010 evaluation campaign. As part of these efforts, we experimented with a number of extensions to the standard phrase-based model that improve performance on the Arabic and Turkish to English translation tasks. We...

READ MORE

Microwave photonic applications of slab-coupled optical waveguide devices

Published in:
2010 23rd Annual Mtg. of the IEEE Photonics Society, 10 November 2010, pp. 479-480.
Topic:

Summary

The semiconductor slab-coupled optical waveguide (SCOW) concept is a versatile device platform that has enabled new classes of high-power, low-noise single-frequency lasers, mode-locked lasers, optical amplifiers, and photodiodes for analog optical links and photonic analog-to-digital converters.
READ LESS

Summary

The semiconductor slab-coupled optical waveguide (SCOW) concept is a versatile device platform that has enabled new classes of high-power, low-noise single-frequency lasers, mode-locked lasers, optical amplifiers, and photodiodes for analog optical links and photonic analog-to-digital converters.

READ MORE

Design, implementation and evaluation of covert channel attacks

Published in:
2010 IEEE Int. Conf. on Technologies for Homeland Security, 8 November 2010, pp. 481-487.

Summary

Covert channel attacks pose a threat to the security of critical infrastructure and key resources (CIKR). To design defenses and countermeasures against this threat, we must understand all classes of covert channel attacks along with their properties. Network-based covert channels have been studied in great detail in previous work, although several other classes of covert channels (hardware based and operating system-based) are largely unexplored. One of our contributions is investigating these classes by designing, implementing, and experimentally evaluating several specific covert channel attacks. We implement and evaluate hardware-based and operating system-based attacks and show significant differences in their properties and mechanisms. We also present channel capacity differences among the various attacks, which span three orders of magnitude. Furthermore, we present the concept of hybrid covert channel attacks which use two or more communication categories to transport data. Hybrid covert channels can be qualitatively harder to detect and counter than traditional covert channels. Finally, we summarize the lessons learned through covert channel attack design and implementation, which have important implications for critical asset protection and risk analysis. The study also facilitates the development of countermeasures to protect CIKR systems against covert channel attacks.
READ LESS

Summary

Covert channel attacks pose a threat to the security of critical infrastructure and key resources (CIKR). To design defenses and countermeasures against this threat, we must understand all classes of covert channel attacks along with their properties. Network-based covert channels have been studied in great detail in previous work, although...

READ MORE

Temporally oblivious anomaly detection on large networks using functional peers

Published in:
IMC'10, Proc. of the ACM SIGCOMM Internet Measurement Conf., 1 November 2010, pp. 465-471.

Summary

Previous methods of network anomaly detection have focused on defining a temporal model of what is "normal," and flagging the "abnormal" activity that does not fit into this pre-trained construct. When monitoring traffic to and from IP addresses on a large network, this problem can become computationally complex, and potentially intractable, as a state model must be maintained for each address. In this paper, we present a method of detecting anomalous network activity without providing any historical context. By exploiting the size of the network along with the minimal overhead of NetFlow data, we are able to model groups of hosts performing similar functions to discover anomalous behavior. As a collection, these anomalies can be further described with a few high-level characterizations and we provide a means for creating and labeling these categories. We demonstrate our method on a very large-scale network consisting of 30 million unique addresses, focusing specifically on traffic related to web servers.
READ LESS

Summary

Previous methods of network anomaly detection have focused on defining a temporal model of what is "normal," and flagging the "abnormal" activity that does not fit into this pre-trained construct. When monitoring traffic to and from IP addresses on a large network, this problem can become computationally complex, and potentially...

READ MORE

Secure channel establishment in disadvantaged networks: optimizing TLS using intercepting proxies

Published in:
MILCOM 2010, IEEE Military Communications Conference , 31 October-3 November 2010.

Summary

Transport Layer Security (TLS) is a secure communication protocol that is used in many secure electronic applications. In order to establish a TLS connection, a client and server engage in a handshake, which usually involves the transmission of digital certificates. In this paper we present a practical speedup of TLS handshakes over bandwidth-constrained, high-latency (i .e. disadvantaged) links by reducing the communication overhead associated with the transmission of digital certificates. This speedup is achieved by deploying two specialized TLS proxies across such links. Working in tandem, one proxy replaces certificate data in packets being sent across the disadvantaged link with a short reference, while the proxy on the other side of the link restores the certificate data in the packet. Local or remote caches supply the certificate data. Our solution preserves the end-to-end security of TLS and is designed to be transparent to third-party applications, and will thus facilitate rapid deployment by removing the need to modify existing installations of TLS clients and TLS servers. Testing shows that this technique can reduce the overall bandwidth used during a handshake by 50% in test emulation and by over 20% of TLS session volume in practice. In addition, it can reduce the time required to establish a secure channel by over 40% across Iridium, a widely used satellite link in practice.
READ LESS

Summary

Transport Layer Security (TLS) is a secure communication protocol that is used in many secure electronic applications. In order to establish a TLS connection, a client and server engage in a handshake, which usually involves the transmission of digital certificates. In this paper we present a practical speedup of TLS...

READ MORE