We propose a neural machine translation architecture that models the surrounding text in addition to the source sentence. These models lead to better performance, both in terms of general translation quality and pronoun prediction, when trained on small corpora, although this improvement largely disappears when trained with a larger corpus. We also discover that attention-based neural machine translation is well suited for pronoun prediction and compares favorably with other approaches that were specifically designed for this task.

In debates over the future of artificial intelligence, many experts think of the new systems as coldly logical and objectively rational. But in a new study, researchers have demonstrated how machines can be reflections of us, their creators, in potentially problematic ways. Common machine learning programs, when trained with ordinary human language available online, can acquire cultural biases embedded in the patterns of wording, the researchers found. These biases range from the morally neutral, like a preference for flowers over insects, to the objectionable views of race and gender. Identifying and addressing possible bias in machine learning will be critically important as we increasingly turn to computers for processing the natural language humans use to communicate, for instance in doing online text searches, image categorization and automated translations.

In debates over the future of artificial intelligence, many experts think of the new systems as coldly logical and objectively rational. But in a new study, researchers have demonstrated how machines can be reflections of us, their creators, in potentially problematic ways. Common machine learning programs, when trained with ordinary human language available online, can acquire cultural biases embedded in the patterns of wording, the researchers found. These biases range from the morally neutral, like a preference for flowers over insects, to the objectionable views of race and gender. Identifying and addressing possible bias in machine learning will be critically important as we increasingly turn to computers for processing the natural language humans use to communicate, for instance in doing online text searches, image categorization and automated translations.

An artificial intelligence tool that has revolutionised the ability of computers to interpret everyday language has been shown to exhibit striking gender and racial biases. The findings raise the spectre of existing social inequalities and prejudices being reinforced in new and unpredictable ways as an increasing number of decisions affecting our everyday lives are ceded to automatons. In the past few years, the ability of programs such as Google Translate to interpret language has improved dramatically. These gains have been thanks to new machine learning techniques and the availability of vast amounts of online text data, on which the algorithms can be trained. However, as machines are getting closer to acquiring human-like language abilities, they are also absorbing the deeply ingrained biases concealed within the patterns of language use, the latest research reveals.

SDL is suing Silicon Valley startup Lilt, alleging patent infringement. In a lawsuit dated April 3, 2017 and filed in the Northern District of California, SDL said Lilt had violated three of its patents and "continues to interfere" with the marketing and sales of SDL products, threatening SDL's relationships with its customers. The patents referred to in the lawsuit are US patents granted Language Weaver, which SDL had acquired in the summer of 2010. Language Weaver was co-founded in 2002 by Daniel Marcu, who eventually joined SDL as Chief Technology Officer post-acquisition. Marcu went on to become the company's Chief Science Officer before moving to Amazon in December 2016 as Director of Machine Translation and Natural Language Processing.

You may want to install Microsoft Translator if you're going to Japan and your vocabulary is limited to "Konnichiwa," "Ohayou" and " Notice me senpai." The app can now turn your spoken words into Nihongo to help you get around the country. Translator can recognize a bevy of languages, but Japanese is only the 10th language its speech translation feature supports. That's right -- it now reads the resulting Japanese words or phrases out loud to make it possible to hold almost real-time conversations with native speakers. The other nine languages in the list are Arabic, Chinese, English, French, German, Italian, Portuguese, Russian and Spanish.

Some giant viruses encode a genome larger than that of some bacteria, but their evolutionary history is a mystery. Examining the genomes within a sample from a wastewater treatment plant in Austria, Schulz et al. assembled a previously undiscovered giant virus genome, which they used to mine genetic databases for related viruses. The authors thus identified a group of giant viruses with more genes encoding components of the protein translation machinery, including aminoacyl transfer RNA synthetases, than in other giant viruses. Phylogenetic analyses suggest that the genes were acquired in an evolutionarily recent time frame, likely from, and as an adaptation to, their hosts.

The only problem with Google's Gboard keyboard for Android is that I'm totally hooked on its best features. Read on for nine of the niftiest Gboard features, from dedicated number rows and an on-demand numeric keypad to "neural" translations and a long-press shortcut for oft-used symbols. Note: Yes, there's also a version of Gboard for iOS, but most of my favorite Gboard tricks only work on the Android version. Tapping a virtual keypad with a single thumb can be something of a stretch if your phone has a massive screen. Luckily, Gboard has a clever feature that makes it easier to tap with just one hand.

Google's latest take on machine translation could make it easier for people to communicate with those speaking a different language, by translating speech directly into text in a language they understand. Machine translation of speech normally works by first converting it into text, then translating that into text in another language. But any error in speech recognition will lead to an error in transcription and a mistake in the translation. Researchers at Google Brain, the tech giant's deep learning research arm, have turned to neural networks to cut out the middle step. By skipping transcription, the approach could potentially allow for more accurate and quicker translations.

The training sentences vary in length from a single word all the way up to very long run-on sentences. TensorFlow does not currently support variable length sequences in RNNs, so this is a problem. In order to deal with these variable length inputs and outputs, the data could be padded to be the same length as the longest input and output, but this would create a lot of pad symbols in our training data and would probably mess with the output. We could also create a model for each possible length of input and output, but that would create a very large number of modules and require a much larger amount of training data. The TensorFlow framework recommends solving this problem through bucketing.