Martin Raab Livres

Automatic translation from one language to another is a highly ambitious task, and there is already a long history of people trying to solve this problem. Yet there is no answer to this problem, but Statistical Machine Translation (SMT) emerged as a promising candidate and is until now of primary research interest. Language Models are very important for SMT, and this book is suggesting and evaluating techniques to improve language models. An excellent source of inspiration for this is the field of speech recognition. The reason is that language models have been studied thoroughly for speech recognition, where language models play a similar role. However, few of the numerous approaches for speech recognition language models have been tested on SMT. Three different language model techniques are evaluated in this book: class base language models, cache language models and sentence mixture language models. Though this book is primarily geared towards SMT, Students and researchers in all areas of language technologies will find a helpful overview of language model techniques in this book.

In theory multiple languages can be recognized just as one language. However, current state of the art speech recognition systems are based on statistical models with many parameters. Extending such models to multiple languages requires more resources. Therefore a lot of research in the area of multilingual speech recognition has proposed techniques to reduce this need for more resources through parameter tying across languages. This work shows that tying at the density level of Hidden Markov Model based speech recognizers offers the greatest flexibility for the design of a multilingual acoustic model. Furthermore, new algorithms are designed and tested for a fast and efficient creation of systems for many different language combinations. These algorithms base on the addition of only relevant Gaussians and on the projection of a Gaussian mixture distribution to new sets of Gaussians. The positive aspects of the architecture proposed in this work are that non-native accent recognition fruitfully applies knowledge about the mother language of the speakers and that an optimal resource allocation for each language can be guaranteed through an online adaptation to the current tasks.

Das derzeitige Steuersystem ermöglicht Anlagestrategien, durch die Steuerzahlungen minimiert werden können. Banken können durch innovative Produkte diese Strategien ermöglichen oder erleichtern und dabei selbst einen Teil des Steuervorteils abschöpfen. Derartige Steuerarbitragestrategien werden in dem vorliegenden Buch systematisch untersucht. Im Rahmen eines State-preference-Ansatzes wird gezeigt, daß es bei dem derzeitigen Einkommensteuersystem wegen der unsystematischen Behandlung von Zins-, Dividendenzahlungen und Wertsteigerungen Arbitragemöglichkeiten geben muß. Der Gesetzgeber kann diese Lücke nur für den Einzelfall aber nicht prinzipiell schließen. Die Banken und Investoren werden dies durch die Konstruktion synthetischer Wertpapiere und komplexer Anlagestrategien nutzen. Die asymmetrische Behandlung der einzelnen Kapitaleinkünfte führt auch zu einer Verzerrung der Unternehmensfinanzierung, die eine Erweiterung des traditionellen Modigliani-Miller-Ansatzes nötig macht.