Neural machine translation aims at building a single large neural network that can be trained to maximize translation performance. The encoder-decoder architecture with an attention mechanism achieves a translation performance comparable to the existing state-of-the-art phrase-based systems on the task of English-to-French translation. However, the use of large vocabulary becomes the bottleneck in both training and improving the performance. In this paper, we propose an efficient architecture to train a deep character-level neural machine translation by introducing a decimator and an interpolator. The decimator is used to sample the source sequence before encoding while the interpolator is used to resample after decoding. Such a deep model has two major advantages. It avoids the large vocabulary issue radically; at the same time, it is much faster and more memory-efficient in training than conventional character-based models. More interestingly, our model is able to translate the misspelled word like human beings.

We propose a general purpose variational inference algorithm that forms a natural counterpart of gradient descent for optimization. Our method iteratively transports a set of particles to match the target distribution, by applying a form of functional gradient descent that minimizes the KL divergence. Empirical studies are performed on various real world models and datasets, on which our method is competitive with existing state-of-the-art methods. The derivation of our method is based on a new theoretical result that connects the derivative of KL divergence under smooth transforms with Stein's identity and a recently proposed kernelized Stein discrepancy, which is of independent interest.

The dominant language models (LMs) such as n-gram and neural network (NN) models represent sentence probabilities in terms of conditionals. In contrast, a new trans-dimensional random field (TRF) LM has been recently introduced to show superior performances, where the whole sentence is modeled as a random field. In this paper, we examine how the TRF models can be interpolated with the NN models, and obtain 12.1\% and 17.9\% relative error rate reductions over 6-gram LMs for English and Chinese speech recognition respectively through log-linear combination.

Efficient index structures for fast approximate nearest neighbor queries are required in many applications such as recommendation systems. In high-dimensional spaces, many conventional methods suffer from excessive usage of memory and slow response times. We propose a method where multiple random projection trees are combined by a novel voting scheme. The key idea is to exploit the redundancy in a large number of candidate sets obtained by independently generated random projections in order to reduce the number of expensive exact distance evaluations. The method is straightforward to implement using sparse projections which leads to a reduced memory footprint and fast index construction. Furthermore, it enables grouping of the required computations into big matrix multiplications, which leads to additional savings due to cache effects and low-level parallelization. We demonstrate by extensive experiments on a wide variety of data sets that the method is faster than existing partitioning tree or hashing based approaches, making it the fastest available technique on high accuracy levels.

Genome-wide association studies (GWAS) offer new opportunities to identify genetic risk factors for Alzheimer's disease (AD). Recently, collaborative efforts across different institutions emerged that enhance the power of many existing techniques on individual institution data. However, a major barrier to collaborative studies of GWAS is that many institutions need to preserve individual data privacy. To address this challenge, we propose a novel distributed framework, termed Local Query Model (LQM) to detect risk SNPs for AD across multiple research institutions. To accelerate the learning process, we propose a Distributed Enhanced Dual Polytope Projection (D-EDPP) screening rule to identify irrelevant features and remove them from the optimization. To the best of our knowledge, this is the first successful run of the computationally intensive model selection procedure to learn a consistent model across different institutions without compromising their privacy while ranking the SNPs that may collectively affect AD. Empirical studies are conducted on 809 subjects with 5.9 million SNP features which are distributed across three individual institutions. D-EDPP achieved a 66-fold speed-up by effectively identifying irrelevant features.

SIRTE, Libya Suicide bombings against Libyan forces battling to oust Islamic State from their former North African stronghold of Sirte killed at least 12 fighters and wounded about 60 there on Thursday, a hospital spokesman said.

LONDON (Thomson Reuters Foundation) - A new technique using artificial intelligence to read satellite images could aid efforts to eradicate global poverty by indicating where help is needed most, a team of U.S. researchers said on Thursday. The method would assist governments and charities trying to fight poverty but lacking precise and reliable information on where poor people are living and what they need, the researchers based at Stanford University in California said. Eradicating extreme poverty, measured as people living on less than 1.25 U.S. a day, by 2030 is among the sustainable development goals adopted by United Nations member states last year. A team of computer scientists and satellite experts created a self-updating world map to locate poverty, said Marshall Burke, assistant professor in Stanford's Department of Earth System Science. It uses a computer algorithm that recognizes signs of poverty through a process called machine learning, a type of artificial intelligence, he said.

And that includes understanding their livelihoods in terms of their sources of income, how their agriculture is performing, how different sectors of the economy perform, and, more specifically, what actually is effective at improving conditions," co-author of the study David Lobell told Mashable in an interview. Therefore, the scientists mapped those dimmer parts of the map at night with daytime photos of the same areas, allowing the computer to pick out patterns -- like road conditions or metal roofs versus thatched roofs -- that indicate a less-developed and possibly poorer region. "The study demonstrates the power of combining multiple data streams to measure things that matter. "The study demonstrates the power of combining multiple data streams to measure things that matter."

The NVIDIA Deep Learning Institute offers self-paced classes for deep learning that feature interactive lectures, hands-on exercises, and live Q&A with instructors. You'll learn everything you need to design, train, and integrate neural network-powered artificial intelligence into your applications with widely used open-source frameworks and NVIDIA software. During the hands-on exercises, you will use GPUs and deep learning software in the cloud. This is an introductory course, so previous experience with deep learning and GPU programming is not required. Please send your questions to DeepLearningInstitute@nvidia.com.

Satellites staring down at Earth can see a lot from their posts in space. Powerful eyes in the sky can pick out homes, natural formations, the pyramids and even small cars driving on roads. And now, scientists are using the wealth of data collected by these satellites to solve major problems on Earth. A new study published in the journal Science this week uses machine learning -- a type of artificial intelligence that lets computer algorithms change when given new data -- coupled with satellite imagery to map poverty in Nigeria, Uganda, Tanzania, Rwanda and Malawi. This new technique could help revolutionize the way groups find impoverished areas and eventually get relief to people living in those specific parts of the world.