Modern Statistical Machine Translation (SMT) systems make their decisions based on multiple information sources, which assess various aspects of the match between a source sentence and its possible translation(s). Tuning a SMT system consists in finding the right balance between these sources so as to produce the best possible output, and is usually achieved through Minimum Error Rate Training ...

Most visual quality inspections in discrete manufacturing are composed of length, surface, angle or intensity measurements. Those are implemented as end-user configurable inspection tools that should not require an image processing expert to set up. Currently available software solutions providing such capability use a flowchart based programming environment, but do not fully address an inspect...

The goal of the EC-funded IST project ORIGAMI was the development of advanced tools and new production techniques for high-quality and seamless integration of real and virtual content in film and TV productions. In particular, the project focused on pre-production tools for automatic virtual set creation through image-based 3-D modelling of environments and objects. One key goal of the project ...

Advances in the miniaturization of electronics is allowing computing devices with various capabilities and interfaces to become part of our daily life. Sensors, actuators, and processing units can now be purchased at very affordable prices. This technology can be networked and used with the coordination of highly intelligent software to understand the events and relevant context of a specific e...

A new risk was born in the mid-1990s known as operational risk. Though its application varied by institutions Basel II for banks and Solvency 2 for insurances the idea stays the same. Firms are interested in operational risk because exposure can be fatal. Hence it has become one of the major risks of the financial sector. In this study, we are going to define operational risk in addition to its...

Statistical machine translation systems are normally optimised for a chosen gain function (metric) by using MERT to find the best model weights. This algorithm suffers from stability problems and cannot scale beyond 20-30 features. We present an alternative algorithm for discriminative training of phrasebasedMT systems, SampleRank, which scales to hundreds of features, equals or beats MERT on b...

| The probability p(k) that the value of a random variable is far away from the mean (e.g. further than k standard deviations away) is so small that this possibility can be often safely ignored. It is desirable to select k for which the dependence of the probability p(k) on the distribution is the smallest possible. Empirically, this dependence is the smallest for k between 1.5 and 2.5. In this...

This paper investigates how inhabitants of collaborative virtual environments (CVEs) can communicate with each other through channels other than speech, and it is primarily concerned with the perception of facial expressions of emotion in CVEs. We outline our experimental work on emotion recognition and investigate to what extent findings from real life socio-psychological research can be appli...

Given the evolution of the Information Technology society, it is now rather simple to acquire products or services in a foreign country. This practice may conduct to the event of conflicts whenever a consumer detects some fault or malfunction in services or products he/she had bought. A situation that may worsen if at the time of the uncovering of the defect, the shopper is already in a differe...

The QMUL submission to IWSLT 2010 is a phrase-based statistical MT system. A multi-stack, multi-beam decoder with several features, with weights tuned on the provided development data through Minimum Error Rate Training (MERT) algorithm. This year QMUL participated in ArabicEnglish, French-English and Turkish-English language pairs of the BTEC task. A discriminative reordering model is added as...